assignement and reflection

687

The Internet of Things as a Platform for Intelligent Applications

LEARNING OBJECTIVES

■■ Describe the IoT and its characteristics ■■ Discuss the benefits and drivers of IoT ■■ Understand how IoT works ■■ Describe sensors and explain their role in IoT applications

■■ Describe typical IoT applications in a diversity of fields

■■ Describe smart appliances and homes ■■ Understand the concept of smart cities, their content, and their benefits

■■ Describe the landscape of autonomous vehicles ■■ Discuss the major issues of IoT implementation

T he Internet of Things (IoT) has been in the technology spotlight since 2014. Its applications are emerging rapidly across many fields in industry, services, govern-ment, and the military (Manyika et al., 2015). It is estimated that 20 to 50 billion “things” will be connected to the Internet by 2020–2025. The IoT connects large numbers of smart things and collects data that are processed by analytics and other intelligent systems. The technology is frequently combined with artificial intelligence (AI) tools for creating smart applications, notably autonomous cars, smart homes, and smart cities.

13.1 Opening Vignette: CNH Industrial Uses the Internet of Things to Excel 688 13.2 Essentials of IoT 689 13.3 Major Benefits and Drivers of IoT 694 13.4 How IoT Works 696 13.5 Sensors and Their Role in IoT 697 13.6 Selected IoT Applications 701 13.7 Smart Homes and Appliances 703 13.8 Smart Cities and Factories 707 13.9 Autonomous (Self-driving) Vehicles 714

13.10 Implementing IoT and Managerial Considerations 717

C H A P T E R

13

688 Part IV • Robotics, Social Networks, AI and IoT

13.1 OPENING VIGNETTE: CNH Industrial Uses the Internet of Things to Excel

CNH Industrial N.V. (CNH) is a Netherlands-based global manufacturer of vehicles for agriculture, construction, and commercial markets. The company produces and services more than 300 types of vehicles and operates in 190 countries where it employs over 65,000 people. The company’s business is continuously growing while operating in a very competitive environment.

THE PROBLEM

To manage and coordinate such a complex business from its corporate office in London, the company needed a superb communication system as well as effective analytical capabili- ties and a customer service network. For example, the availability of repair parts is critical. Customers’ equipment does not work until a broken part is replaced. Competitive pressures are very strong, especially in the agriculture sector where weather conditions, seasonality, and harvesting pressure may complicate operations. Monitoring and controlling equipment properly is an important competitive factor. Predicting equipment failures is very desirable. Rapid connectivity with customers and the equipment they purchase from CNH is essential as are efficient data monitoring and data collection. Both CNH and its customers need to make continuous decisions for which real-time flow of information and communication is essential.

THE SOLUTION

Using PTC Transformational Inc. as an IoT, vendor, CNH implemented an IoT-based sys- tem with internal structural transformation in order to solve its problems and reshape its connected industrial vehicles. The initial implementation was in the agricultural sector. The details of the implementation are provided by PTC, Inc. (2015). The highlights of this IoT are summarized next.

• Connects all vehicles (those that are equipped with sensors and are connected to the system) in hundreds of locations worldwide to CNH’s command and control center. This connection enables monitoring performance.

• Monitors the products’ condition and operation as well as their surrounding envi- ronments through sensors. It also collects external data, such as weather conditions.

• Enables customization of products’ performance at customers’ sites. • Provides the data necessary for optimizing the equipment’s operation. • Analyzes the performance of the people who drive CNH’s manufactured vehicles

and recommends changes that can improve the vehicles’ efficiency. • Predicts the range of the fuel supply in the vehicles. • Alerts owners to the needs (and timing) of preventive maintenance (e.g., by moni-

toring usage and/or predicting failures) and orders the necessary parts for such service. This enables proactive and preventive maintenance practices.

• Finds when trucks are overloaded (too much weight), violating CNH’s warranty. • Provides fast diagnosis of products’ failures. • Enables the delivery of trucks on schedule by connecting them to planners and

with delivery sources and destinations. • Helps farmers to optimally plan the entire farming cycle from preparing the soil to

harvesting (by analyzing the weather conditions). • Analyzes collected data and compares them to standards.

All of this is done mostly wirelessly.

Chapter 13 • The Internet of Things as a Platform for Intelligent Applications 689

THE RESULTS

According to Marcus (2015), CNH halved the downtime of its participating equipment at customer sites by using the IoT. Parts for incoming orders can be shipped very quickly. IoT use also helped farmers monitor their fields and equipment to improve efficiency. The company is now showing customers less effective examples of operations and superb operating practices. In addition, product development benefits from the analysis of col- lected data.

Sources: Compiled from PTC, Inc. (2015), Marcus (2015), and cnhindustrial.com/en-us/pages/homepage.aspx.

u QUESTIONS FOR THE OPENING VIGNETTE

1. Why is the IoT the only viable solution to CNH’s problems? 2. List and discuss the major benefits of IoT. 3. How can CNH’s product development benefit from the collected data about usage? 4. It is said that the IoT enables telematics and connected vehicles. Explain. 5. Why is IoT considered the “core of the future business strategy”? 6. It is said that the IoT will enable new services for CNH (e.g., for sales and

collaboration with partners). Elaborate.

7. View Figure 13.1 (The process of IoT) and relate it to the use of IoT at CNH. 8. Identify decision support possibilities. 9. Which decisions made by the company and its customers are supported by IoT?

WHAT WE CAN LEARN FROM THIS VIGNETTE

First, we learned how IoT provides an infrastructure for new types of applications that connect thousands of items to a decision-making center.

Second, we learned about the flow of data collected by sensors from vehicles and the environment around them and their transmittal for analytical processing.

Third, the manufacturer of the vehicles and their owners and users can enjoy tremen- dous benefits from using the system.

Finally, this, IoT provides an efficient communication and collaboration framework for decision makers, the manufacturer’s organization, and the users of the purchased equipment.

In this chapter, we elaborate on the technologies involved and the process of the IoT operation. We also describe its major application in enterprises, homes, smart cities, and autonomous (smart) vehicles.

13.2 ESSENTIALS OF IoT

The Internet of Things (IoT) is an evolving term with several definitions. In general, IoT refers to a computerized network that connects many objects (people, animals, de- vices, sensors, buildings, items) each with an embedded microprocessor. The objects are connected, mostly wirelessly, to the Internet forming the IoT. The IoT can exchange data and allow communication among the objects and with their environments. That is, the IoT allows people and things to be interconnected anytime and anyplace. Embedded sensors that collect and exchange data make up a major portion of the objects and the IoT. That is, IoT uses ubiquitous computing. Analysts predict that by the year 2025, more than 50 billion devices (objects) will be connected to the Internet, creating the backbone

690 Part IV • Robotics, Social Networks, AI and IoT

of IoT applications. The challenges and opportunities of this disruptive technology (e.g., for cutting costs, creating new business models, improving quality) are discussed in an in- terview with Peter Utzschneider, vice president of product management for Java at Oracle (see Kvitka, 2014). In addition, you can join the conversations at iotcommunity.com. For Intel’s vision of a fully connected world, see Murray (2016).

Embedding computers and other devices that can be switched on and off into ac- tive items anywhere and connecting all devices to the Internet (and/or to each other) permit extensive communication and collaboration between users and items. By con- necting many devices that can talk to each other, one can create applications with new functionalities, increase the productivity of existing systems, and drive the benefits dis- cussed later. This kind of interaction opens the door to many applications. For business applications of the Internet of Things, see Jamthe (2016). In addition, check the “Internet of Things Consortium” (iofthings.org) and its annual conferences. For an infographic and a guide, see intel.com/content/www/us/en/internet-of-things/infographics/ guide-to-iot.html.

Definitions and Characteristics

There are several definitions of IoT. Kevin Ashton, who is credited with the term the “Internet of Things,” provided the

following definition: “The Internet of Things means sensors connected to the Internet and behaving in an Internet-like way by making open, ad hoc connections, sharing data freely, and allowing unexpected applications, so computers can understand the world around them and become humanity’s nervous system” (term delivered first in a 1999 oral presentation. See Ashton, 2015).

Our working definition is: The IoT is a network of connected computing devices including different types of

objects (e.g., digital machines). Each object in the network has a unique identifier (UID), and it is capable of collecting and transferring data automatically across the network.

The collected data has no value until it is analyzed, as illustrated in the opening vignette.

Note that the IoT allows people and things to interact and communicate at any time, any place, regarding any business topic or service.

According to Miller (2015), the IoT is a connected network in which:

• Large numbers of objects (things) can be connected. • Each thing has a unique definition (IP address). • Each thing has the ability to receive, send, and store data automatically. • Each thing is delivered mostly over the wireless Internet. • Each thing is built upon machine-to-machine (M2M) communication.

Note that, in contrast with the regular Internet that connects people to each other using computing technology, the IoT connects “things” (physical devices and people) to each other and to sensors that collect data. In Section 13.4, we explain the process of IoT.

SIMPLE EXAMPLES A common example of the IoT is the autonomous vehicle (Section 13.9). To drive on its own, a vehicle needs to have enough sensors that automatically monitor the situation around the car and take appropriate actions whenever necessary to adjust any set- ting, including the car’s speed, direction, and so on. Another example that illustrates the IoT phenomenon is the company Smartbin. It has developed trash containers that include sen- sors to detect their fill levels. The trash collection company is automatically notified to empty a trash container when the sensor detects that the bin has reached the fill level.

Chapter 13 • The Internet of Things as a Platform for Intelligent Applications 691

A common example people give to illustrate IoT is the idea that a refrigerator could automatically order food (e.g., milk) when it detects that the food has run out! Clorox introduced a new Brita filter so that a Wi-Fi–enabled mechanism can order water filters by itself when it detects that it is time to change them. In these examples, a human does not have to communicate with another human or even with a machine.

IoT IS CHANGING EVERYTHING According to McCafferty (2015), the IoT is changing everything. This has been verified by a 2016 survey reported by Burt (2016). For how manufacturing is revolutionized by IoT, see Greengard (2016). Here are a few examples that he provided:

• “Real-time systems make it possible to know where anyone is at any moment, which is helpful to secured locations as military bases and seeking to push promo- tions to consumers.”

• “Fleet tracking systems allow logistics and transport firms to optimize routing, track vehicle speeds and locations, and analyze driver and route efficiencies.”

• “Owners and operators of jet engines, trains, factory equipment, bridges, tunnels, etc., can stay ahead of repairs through machines that monitor for preventive main- tenance.” (opening case)

• “Manufacturers of foods, pharmaceuticals and other products monitor temperature, humidity and other variables to manage quality control, receiving instant alerts when something goes wrong.”

These changes are facilitated by AI systems, which enhance analytics and automate or support decision making.

The IoT Ecosystem

When billions of things are connected to the Internet with all the supporting services and connected IT infrastructure, we can see a giant complex, which can be viewed as a huge ecosystem. The Internet of Things ecosystem refers to all components that en- able users to create IoT applications. These components include gateways, analytics, AI algorithms, servers, data storage, security, and connectivity devices. A pictorial view is provided in Figure 13.1 in which applications are shown on the left side and the building blocks and platforms on the right side. An example of an IoT application is provided in the opening vignette. It illustrates a network of sensors that collects information, which is transmitted to a central place for processing and eventually for decision support. Thus, the IoT applications are subsets of the IoT ecosystem.

A basic discussion, terms, major companies, and platforms is provided by Meola (2018).

Structure of IoT Systems

Things in IoT refers to a variety of objects and devices ranging from cars and home appliances to medical devices, computers, fitness tracers, hardware, software, data, sensors, and much more. Connecting things and allowing them to communicate is a necessary capability of an IoT application; but for more sophisticated applications, we need additional components: a control system and a business model. The IoT en- ables the things to sense or be sensed wirelessly across the network. A non-Internet example is a temperature control system in a room. Another non-Internet example is a traffic signal at intersections of roads where camera sensors recognize the cars com- ing from each direction and a control system adjusts the time for changing the lights according to programmed rules. Later, we will introduce the reader to many Internet- based applications.

692 Part IV • Robotics, Social Networks, AI and IoT

IoT TECHNOLOGY INFRASTRUCTURE From a bird’s-eye view, IoT technology can be divided into four major blocks. Figure 13.2 illustrates them.

1. Hardware: This includes the physical devices, sensors, and actuators where data are produced and recorded. The devices are the equipment that needs to be con- trolled, monitored, or tracked. IoT sensor devices could contain a processor or any computing device that parses incoming data.

Internet of Things

2016

Platforms & Enablement (Horizontals)

B u ild

in g B

lo c ks

Applications

(Verticals)

W earables

Fitness HealthEntertainmentFamily

Sports

Elderly

Toys

Automation

Hubs

Securit y

Kitch en

Sen sing

Co nsu

me r

Ro bo

tic s

Pe ts

Ga rd

en

Tr ac

ke rs

A ut

om ob

ile s

A ut

on om

ou s

U A V s

S pa

ce B ic yc

le s/

M ot

or bi

ke s

H e a lt h ca

re

R e ta

il

S m

a rt O

ffi c e

A g ric

u ltu

re In

fra stru

ctu re

M achines

E nergy

S upply C

hain R obotics

Industrial W earables

Softw are

Full Stack

Connectivity

Sensor Networks

Developer

Security

Analytics

Open Source

Virtual Reality

Augme nted R

eality Oth

er

Pri ntin

g/S can

nin g

Co nt

en t/D

es ign

Pr oc

es so

rs /C

hi ps

S en

so rs

P ar

ts /K

it s

C ha

rg in

g

P ro

to co

ls

T e le

c o m

M 2 M

W i- F iC

lo u d

M o b il O

S

C onsultants/S

ervices

Incubators

A lliances

Funding

Home

V eh

ic le s

E n te

rp ri s e

Industrial

Internet

Platforms

Interfaces

3D

H ar

dw ar

e

C o n n e c ti vi

tyS o ftw

a re

P artners

Personal P

a ym

e n ts

/L o ya

lt y

FIGURE 13.1 The IoT 2016 (Ecosystem).

Chapter 13 • The Internet of Things as a Platform for Intelligent Applications 693

2. Connectivity: There should be a base station or hub that collects data from the sensor-laden objects and sends those data to the “cloud” to be analyzed. Devices are connected to a network to communicate with other networks or other applications. These may be directly connected to the Internet. A gateway enables devices that are not directly connected to the Internet to reach the cloud platform.

3. Software backend: In this layer, the data collected are managed. Software back- end manages connected networks and devices and provides data integration. This may very well be in the cloud.

4. Applications: In this part of IoT, data are turned into meaningful information. Many of the applications can run on smartphones, tablets, and PCs and do some- thing useful with the data. Other applications can run on the server and provide results or alerts through dashboards or messages to the stakeholders.

To assist with the construction of IoT systems, one may use IoT platforms. For informa- tion, see Meola (2018).

IoT Devices Applications

Internet Network

Data Storage Validate

Built

Test

Analytics

D ata

Re qu

es t

An al ys

is

AnalysisData

Cloud-Based Storage & Computing

Gateway

FIGURE 13.2 The Building Blocks of IoT.

694 Part IV • Robotics, Social Networks, AI and IoT

IoT PLATFORMS Because IoT is still evolving, many domain-specific and application- specific technology platforms are also evolving. Not surprisingly, many of the major vendors of IoT platforms are the same ones who provide analytics and data storage services for other application domains. These include Amazon AWS IoT, Microsoft Azure IoT suite, Predix IoT Platform by General Electric (GE), and IBM Watson IoT platform (ibm.com/us-en/marketplace/internet-of-things-cloud). Teradata Unified Data Architecture has similarly been applied by many customers in the IoT domain.

u SECTION 13.2 REVIEW QUESTIONS

1. What is IoT? 2. List the major characteristics of IoT. 3. Why is IoT important? 4. List some changes introduced by IoT. 5. What is the IoT ecosystem? 6. What are the major components of an IoT technology?

13.3 MAJOR BENEFITS AND DRIVERS OF IoT

The major objective of IoT systems is to improve productivity, quality, speed, and the quality of life. There are potentially several major benefits from IoT, especially when combined with AI, as illustrated in the opening case. For a discussion and examples, see Jamthe, 2015.

Major Benefits of IoT

The following are the major benefits of IoT:

• Reduces cost by automating processes. • Improves workers’ productivity. • Creates new revenue streams. • Optimizes asset utilization (e.g., see the opening vignette). • Improves sustainability. • Changes and improves everything. • May anticipate our needs (predictions). • Enables insights into broad environments (sensors collect data). • Enables smarter decisions/purchases. • Provides increased accuracy of predictions. • Identifies problems quickly (even before they occur). • Provides instant information generation and dissemination. • Offers quick and inexpensive tracking of activities. • Makes business processes more efficient. • Enables communication between consumers and financial institutions. • Facilitates growth strategy. • Fundamentally improves the use of analytics (see the opening vignette). • Enables better decision making based on real-time information. • Expedites problem resolution and malfunction recovery. • Supports facility integration. • Provides better knowledge about customers for personalized services and marketing.

Chapter 13 • The Internet of Things as a Platform for Intelligent Applications 695

Major Drivers of IoT

The following are the major drivers of IoT:

• The number of “things”—20 to 50 billion—may be connected to the Internet by 2020–2025.

• Connected autonomous “things”/systems (e.g., robots, cars) create new IoT applications.

• Broadband Internet is more widely available, increasing with time. • The cost of devices and sensors is continuously declining. • The cost of connecting the devices is decreasing. • Additional devices are created (via innovations) and are interconnected easily

(e.g., see Fenwick, 2016). • More sensors are built into devices. • Smartphones’ penetration is skyrocketing. • The availability of wearable devices is increasing. • The speed of moving data is increasing to 60 THz. • Protocols are developing for IoT (e.g., WiGig). • Customer expectations are rising; innovative customer services are becoming a

necessity. • The availability of IoT tools and platforms is increasing. • The availability of powerful analytics that are used with IoT is increasing.

Opportunities

The benefits and drivers just listed create many opportunities for organizations to excel in the economy (e.g., Sinclair, 2017), in many industries and in different settings.

McKinsey Global Institute (Manyika et al., 2015) provides a comprehensive list of settings where IoT is or can be used with examples in each setting. A 2017 study (Staff, 2017) revealed a dramatic increase in the capabilities and benefits of IoT.

HOW BIG CAN AN IoT NETWORK BE? While there will be billions of things connected to the Internet soon, not all of them will be connected in one IoT network. However, an IoT network can be very large, as we show next.

Example: World’s Largest IoT Is Being Built in India (2017)

This network is being constructed by Tata Communications of India and HP Enterprises (HPE) of the United States, over the HPE Universal IoT Platform. The things to be con- nected exist in 2,000 communities and include computing devices, applications, and IoT solutions, connected over the Lo Ra network, a wireless communication protocol for wide area networks. The things are in smart buildings, utilities, university campuses, se- curity systems, vehicles and fleets, and healthcare facilities.

The project is to be implemented in phases with proof-of-concept applications to be tested first. The network will bring services to 400 million people. For details, see Shah (2017).

u SECTION 13.3 REVIEW QUESTIONS

1. List the benefits of IoT for enterprises. 2. List the benefits of IoT for consumers. 3. List the benefits of IoT for decision making. 4. List the major drivers of IoT.

696 Part IV • Robotics, Social Networks, AI and IoT

13.4 HOW IoT WORKS

IoT is not an application. It is an infrastructure, platform, or framework that is used to support applications. The following is a comprehensive process for IoT applications. In many cases, IoT follows only portions of this process.

The process is explained in Figure 13.3. The Internet ecosystem (top of the figure) includes a large number of things. Sensors and other devices collect information from the ecosystem. The collected information can be displayed, stored, and processed analyti- cally (e.g., by data mining). This analysis converts the information into knowledge and/or intelligence. Expert systems or machine learning may help in turning the knowledge into decision support (made by people and/or machines), which is evidenced by improved actions and results.

The generated decisions can help in creating innovative applications, new business models, and improvements in business processes These result in “actions,” which may impact the original scenario or other things. The opening vignette illustrates this process.

Note that most of the existing applications are in the upper part of the figure, which is called sensor to insight, meaning up to the creation of knowledge or to the delivery of new information. However, now, the focus is moving to the entire cycle (i.e., sensor to action).

The IoT may generate a huge amount of data (Big Data) that needs to be analyzed by various business intelligence methods, including deep learning, or advanced AI methods.

IoT and Decision Support

As stated earlier, the IoT creates knowledge and/or intelligence, which is submitted as support to decision makers or is inputted to automated decision support entities. The transition from data collection to decision support may not be simple due to the large amount of data, some of which are irrelevant. Large-scale IoT usually needs to filter the

Sensors

Information flow

Collected Stored Transferred

Wireless Systems

Analysis, Mining, Processing Intelligence, Knowledge

Machine learning

Decision making

Innovation New business model

Improvements Actions

Other ‘things’, other systems

‘Things’

Internet

People and/or machines

The Internet Ecosystem Wireless

iPaq Wireless Laptop

Wireless Desktop

Wireless Print Server

Router/Wireless Access Point

Wired Laptop

Wired Desktop

Cable/DSL Modem

INTERNET

FIGURE 13.3 The Process of IoT.

Chapter 13 • The Internet of Things as a Platform for Intelligent Applications 697

collected data and “clean” them before they can be used for decision support, particularly if they are used as a base for automated decision making.

u SECTION 13.4 REVIEW QUESTIONS

1. Describe the major components of IoT. 2. Explain how the IoT works following the process illustrated in Figure 13.3. 3. How does IoT support decision making?

13.5 SENSORS AND THEIR ROLE IN IoT

As illustrated in the opening vignette to this chapter, sensors play a major role in IoT by collecting data about the performance of the things that are connected to the Internet and monitoring the surrounding environment, collecting data there too if necessary. Sensors can transmit data and sometimes even process it prior to transmission.

Brief Introduction to Sensor Technology

A sensor is an electronic device that automatically collects data about events or changes in its environment. Many IoT applications include sensors (see the opening vignette). The collected data are sent to other electronic devices for processing. There are several types of sensors and several methods for collecting data. Sensors often generate signals that are converted to human-readable displays. In addition to their use in IoT, sensors are essen- tial components in robotics and autonomous vehicles. Each sensor usually has a limit on the maximum distance that it can detect (nominal range). Sensors of a very short range known as proximity sensors are more reliable than those that operate in larger ranges. Each IoT network may have millions of sensors. Let us see how sensors work with IoT in Application Case 13.1.

The Problem

Over 20 million passengers use the airport annually, and their number increases by more than 10 percent every year. Obviously, the number of flights is large and also increasing annually. The growth increases air pollution as well. The airport has a strong com- mitment to environmental protection, so manage- ment has looked for an environmental control solu- tion. The objective was to make the airport carbon neutral. The large number of planes in the air and on the ground and the fact that airplanes frequently move require advanced technologies for the solution.

The Solution

A reasonable way to deal with moving airplanes was to use IoT, a technology that when combined

with AI-based sensors enables environmental moni- toring, analysis, and reporting, all of which provide the background information for decisions regarding minimizing the air pollution.

Two companies combined their expertise for this project: EXM of Greece, which specializes in IoT prediction analytics and innovative IoT solutions, and Libelium of the United States, which specializes in AI-related sensors, including those for environmental use. The objective of the project was to properly mon- itor air quality inside and outside the airport and to identify, in real time, the aircraft location on the ground and to take corrective actions whenever needed.

Ad Hoc Air Quality Monitoring and Analysis

The airport now has an air quality monitoring net- work. The solution includes Libelium’s sensor

Application Case 13.1 Using Sensors, IoT, and AI for Environmental Control at the Athens, Greece, International Airport

(Continued )

698 Part IV • Robotics, Social Networks, AI and IoT

platform connected in a cost-effective manner. The different sensors measure temperature, humidity, atmosphere pressure, ozone level, and particulate matter. The readings of the sensors are transmitted to IoT for reporting and then analysis. The sensors were improved by using AI features. Therefore, their accuracy increased. In addition, security and energy consumption are also being controlled.

Aircraft Location at the Airport

To identify the exact location of the aircrafts dur- ing takeoff and landing, the project uses acoustic measurement mechanisms. This is accomplished by using noise sensors placed in different locations. The sensors measure real-time noise level, which is evaluated by analytics. Overall, the system provides a noninvasive IoT solution.

Placement of sensors was difficult due to safety, security, and regulation considerations. Therefore, the sound monitoring subsystem had to be self-managed (autonomous), bearing solar panels and batteries that provided the electricity. In addi- tion, the system utilizes a dual wireless communica- tion system (known as GPPS).

The collected noise data are correlated with types of airplane and flights at the IoT backend. All data are analyzed by the airport environmen- tal department and used for decisions regarding improvements of pollution control.

Technology Support

The solution combines an IoT system with AI-based analytics, visualization, and reporting and is exe- cuted in the cloud. In addition, the system has on- site sensors and communication infrastructures. Low-power wireless sensors monitor water and gas consumption indoors as well as air quality in the parking sites. Vendors’ products, such as Microsoft Azure and IBM Bluemix, support the project and provide the necessary flexibility.

Sources: Compiled from Hedge (2017) and Twentyman (2017).

Questions for Case 13.1

1. What is the role of IoT in the project?

2. What is the role of sensors?

3. What are the benefits of the project?

How Sensors Work with IoT

In large-scale applications, sensors collect data that are transferred to processing in the “cloud.” Several platforms are used for this process as discussed in Application Case 13.2.

Rockwell Automation is one of the world’s largest providers of industrial automation and information solutions. It has customers in more than 80 countries worldwide and around 22,500 employees. One of its business areas of focus is assisting oil and gas com- panies in exploration. An example is Hilcorp Energy, a customer company that drills oil in Alaska. The equipment used in drilling, extracting, and refining oil is very expensive. A single fault in the equipment can cost the company around $100,000 to $300,000 per day in lost production. To deal with this

problem, it needed technology to monitor the status of such piece of equipment remotely and to predict failures that are likely to happen in the future.

Rockwell Automation considered the oppor- tunity to expand its business in oil and gas indus- tries by gathering data from the exploration sites and analyzing them to improve preventive main- tenance decision making regarding the critical equipment, thus, minimizing downtime and drive better performance. The company utilizes its vision of Connected Enterprise with Microsoft’s software

Application Case 13.2 Rockwell Automation Monitors Expensive Oil and Gas Exploration Assets to Predict Failures

Application Case 13.1 (Continued)

Chapter 13 • The Internet of Things as a Platform for Intelligent Applications 699

Sensor Applications and Radio-Frequency Identification (RFID) Sensors

There are many types of sensors. Some measure temperature; others measure humid- ity. Many sensors collect information and transmit it as well. For a list of 50 sensor applications with a large number of related articles, see libelium.com/resources/ top_50_iot_sensor_applications_ranking/.

A well-known type of sensor that plays an important role in IoT is radio-frequency identification.

RFID SENSORS Radio-frequency identification (RFID) is part of a broader ecosys- tem of data capture technologies. Several forms of RFID in conjunction with other sen- sors play a major role in IoT applications. Let us see first what RFID is, as discussed in Technology Insights 13.1.

to monitor and support oil and gas equipment placed in remote areas. Rockwell is now provid- ing solutions to predict failure of equipment along the entire petroleum supply chain, monitoring its health and performance in real time, and to prevent failures in the future. Solutions are provided in the following areas.

• Drilling: Hilcorp Energy has its pump- ing equipment stationed in Alaska where it drills for oil 24 hours a day. A single failure in equipment can cost Hilcorp a large amount of money. Rockwell connected electrical variable drives of pumping equipment to be processed in the “cloud,” to control its machines thou- sands of miles away from the control room in Ohio. Sensors capture data, and through Rock- well’s control gateway, these data are passed to Microsoft Azure Cloud. The solutions derived reach Hilcorp engineers through digital dash- boards that provide real-time information about pressure, temperature, flow rate, and dozens of other parameters that help engineers moni- tor the equipment’s health and performance. These dashboards also display alerts about any possible issues. When one of Hilcorp’s pieces of pumping equipment failed, it was identified, tracked, and repaired in less than an hour, sav- ing six hours of tracing the failure and the large cost of lost production.

• Building smarter gas pumps: Today, some delivery trucks use liquid natural gas (LNG) as fuel. Oil companies are updating their fill- ing stations to incorporate LNG pumps. Rock-

well Automation installed sensors and variable frequency drives at these pumps to collect real- time data about equipment operations, fuel in- ventory, and consumption rate. This data are transmitted to Rockwell’s cloud platform for processing. Rockwell then generates interac- tive dashboards and reports using Microsoft Azure (an IoT platform). Results are forwarded to the appropriate stakeholders, giving them a good idea about the health of their capital assets.

The Connected Enterprise solution by Rockwell has accelerated growth for many oil and gas companies like Hilcorp Energy by bringing their operations data to the cloud platform and helping them reduce costly downtime and maintenance. It has resulted in a new business opportunity for industrial age stal- warts like Rockwell Automation.

Sources: customers.microsoft.com (2015); Rockwell Automation: Fueling the Oil and Gas Industry with IoT; https://customers. microsoft.com/Pages/CustomerStory.aspx?recid=19922; Microsoft.com. (n.d.). “Customer Stories | Rockwell Automation,” https://www.microsoft.com/en-us/cloud-platform/ customer-stories-rockwell-automation (accessed April 2018).

Questions for Case 13.2

1. What type of information would likely be col- lected by an oil and gas drilling platform?

2. Does this application fit the three V’s (volume, variety, velocity) of Big Data? Why or why not?

3. Which other industries (list five) could use simi- lar operational measurements and dashboards?

700 Part IV • Robotics, Social Networks, AI and IoT

TECHNOLOGY INSIGHTS 13.1 RFID Sensors

RFID is a generic technology that refers to the use of radio-frequency waves to identify objects. Fundamentally, RFID is one example of a family of automatic identification technologies that also in- cludes ubiquitous barcodes and magnetic strips. Since the mid-1970s, the retail supply chain (among many other areas) has used barcodes as the primary form of automatic identification. RFIDs can store a much larger amount of data than barcodes. Also, they can be accessed from a longer distance wire- lessly. These potential advantages of RFID have prompted many companies (led by large retailers such as Walmart and Target) to aggressively pursue it as a way to improve their supply chains and thus reduce costs and increase sales. For details, see Chapter 8 in Sharda et al. (2018).

How does an RFID work? In its simplest form, an RFID system consists of a tag (attached to the product to be identified), an interrogator (i.e., RFID reader), one or several antennae at- tached to the reader, and a computer program (to control the reader and capture the data). At present, the retail supply chain has primarily been interested in using passive RFID tags. Passive tags receive energy from the electromagnetic field created by the interrogator (e.g., a reader) and backscatter information only when it is requested. The passive tag remains energized only while it is within the interrogator’s magnetic field.

In contrast, active tags have a battery to energize themselves. Because active tags have their own power source, they do not need a reader to energize them; instead, they can initiate the data transmission process on their own. As compared to passive tags, active tags have a longer read range, better accuracy, more complex rewritable information storage, and richer processing capabilities. On the negative side, their batteries cause active tags to have a limited life span, be larger in size than passive tags, and be more expensive. Currently, most retail applications are de- signed and operated with passive tags, each of which costs only a few cents. Active tags are most frequently found in defense and military systems, yet they also appear in technologies such as EZ Pass whose tags (called transponders) are linked to a prepaid account that, for example, enables drivers to pay tolls later, by driving past a reader rather than stopping to pay at a tollbooth.

Note: There are also semipassive tags with limited active tag capabilities.

The most commonly used data representation for RFID technology is the Electronic Product Code (EPC), which is viewed by many in the industry as the next generation of the Universal Product Code (UPC), most often represented by a barcode. Like the UPC, the EPC consists of a series of numbers that identifies product types and manufacturers across the supply chain. The EPC also includes an extra set of digits to uniquely identify items.

Use of RFID and Smart Sensors in IoT Basic RFID tags, either active or passive, are not sensors. The purpose of the tags is to identify objects and determine their location (e.g., for the purpose of counting objects). To make them useful for most IoT applications, the tags need to be upgraded (e.g., by adding on-board sen- sors). These RFIDs called RFID sensors have more capabilities than RFID tags, or basic sensors. For a detailed discussion about the role of RFID in the IoT, see Donaldson (2017).

RFID sensors are wireless sensors that communicate, via mash networks or conventional RFID readers, and they include identifiable ID. The RFID reader sends token information into gateways, such as AWS IoT service. This confirmation can be processed, resulting in some action.

SMART SENSORS AND IoT There are several types of smart sensors with different levels of capabilities when integrated into IoT. A smart sensor is one that senses the environment and processes the input it collects by using its built-in computing capabilities (e.g., a micro- processing). The processing is preprogrammed. Results are passed on. Depending on the internal computing quality, smart sensors can be more automated and accurate than other sensors and can filter out unwanted noise and compensate for errors before sending the data.

Smart sensors are crucial and an integral element in the IoT. They can include special components, such as amplifiers, analog filters, and transducers, to support IoT.

Chapter 13 • The Internet of Things as a Platform for Intelligent Applications 701

In addition, smart sensors for IoT can include special software for data conversion, digital processing, and communication capability to external devices.

According to a major study (Burkacky et al., 2018), sensors are getting smarter. Those on vehicles are examples. Vehicles can make the transition from being a hardware- driven machine to being a software-driven electronic device. Software can cost over 35 percent of the cost of vehicle production.

For further information, see Scannell (2017), Gemelli (2017), and Technavio (2017).

u SECTION 13.5 REVIEW QUESTIONS

1. Define sensor. 2. Describe the role of sensors in IoT. 3. What is RFID? What is a RFID sensor? 4. What role does the RFID perform in IoT? 5. Define smart sensor and describe its role in IoT.

13.6 SELECTED IoT APPLICATIONS

We start with a well-known example: Imagine that your refrigerator can sense the amount of food in it and send you a text message when inventory is low (sensor-to-insight in Figure 13.3). One day refrigerators will also be able to place an order for items that need replenishment, pay for them, and arrange delivery (sensor-to-action). Let us look at some other, less futuristic enterprise applications.

A Large-scale IoT in Action

Existing contribution of IoT has centered on large organizations.

Example French National Railway System’s Use of IoT

SNCF, the French national railway system, uses IoT to provide quality, availability, and safety for its nearly 14 million passengers. The company sncf.com improved its operations using IoT (Estopace, 2017a). To manage 15,000 trains and 30,000 kilometers of tracks is not simple, but IBM Watson, using IoT and analytics, helped to do just that. Thousands of sensors that are in- stalled on the trains, tracks, and train stations gather data that Watson processes. In addition, all business process operations were digitized to fit into the system. Information concerning pos- sible cyberattacks was also programmed into the system. All collected Big Data were prepared for decision support. IBM Watson’s platform is scaleable and can handle future expansions.

To understand the magnitude of this IoT network, consider that the mass transit lines in Paris alone required 2,000 sensors forwarding information from more than 7,000 data points each month. The systems enable engineers to remotely monitor 200 trains at a time for any mechanical and electrical operations and malfunctions while trains are mov- ing. In addition, by using a predictive analytic model, the company can schedule preven- tive maintenance to minimize failures. Therefore, if you are one of the train travelers, you can relax and enjoy your trip.

Examples of Other Existing Applications

The following examples of the use of IoT applications are based on information from Koufopoulos (2015):

• Hilton Hotel. Guests can check in directly to their rooms with their smartphones (no check-in lobby is needed, no keys are used). Other hotel chains follow suit.

• Ford. Users can connect to apps by voice. Autopaying for gas and preordering drinks at Starbucks directly from Ford’s cars are in development.

702 Part IV • Robotics, Social Networks, AI and IoT

• Tesla. Tesla’s software autonomously schedules a valet to pick up a car and drive it to Tesla’s facility when a car needs repair or schedule service. Tesla trucks, man- aged by IoT, will be driverless one day.

• Johnnie Walker. The whiskey company connected 100,000 of its bottles to the Internet for Brazil’s Father’s Day. Using smart labeling, buyers can create personal- ized videos to share with their fathers on social networks. Fathers also get promo- tions to buy more whiskey if they like it.

• Apple. Apple enables users of iPhones, Apple Watches, and Home kits to stream- line shopping with Apple Pay.

• Starbucks Clover Net in the Cloud. This system connects coffee brewers to customers’ preferences. It also monitors employee’s performance, improves recipes, tracks consumption patterns, and so on.

A large number of consumer applications of IoT is reported by Jamthe (2016) and Miller (2015). For a list of IoT applications related to IBM Watson, see ibm.com/ internet-of-things/.

Many companies are experimenting with IoT products for retailing (business to consumer, or B2C) and business to business (B2B) in areas such as operations,  transportation, logistics, and factory warehousing. For the approaches of Apple and Amazon, see appadvice.com/post/apple-amazons-smart-home- race/736365/.

Note: For many case studies and examples of the IoT, see ptc.com/en/ product-lifecycle-report/services- and-customer-success-collide-in-the-iot, divante.co/blog/internet-e-commerce, and Greengard (2016). IoT is also used for many applications inside enterprises (see McLellan, 2017a), and military purposes (see Bordo, 2016).

HOW IoT IS DRIVING MARKETING According to Durrios (2017), IoT can drive marketing opportunities in the following four ways:

1. Disruptive data collection. IoT collects more data about customers from more data sources than other technologies do. This includes data from wearables, smart homes, and everything consumers do. In addition, IoT provides data about changes in con- sumer preferences and behavior.

2. Real-time personalization. IoT can provide more accurate information about spe- cific customers buying decisions, for example. IoT can identify customer expecta- tions and direct customers to specific brands.

3. Environmental attribution. IoT can monitor environments regarding ad delivery for specific places, customers, methods, and campaigns. IoT can facilitate research of business environment; factors such competition, pricing, weather conditions, and new government regulations are observed.

4. Complete conversation path. IoT initiatives expand and enrich the digital channel of conversations between customers and vendors, especially those using wireless digital engagement. IoT also provides insight on consumer purchasing paths. In addition, marketers will receive improved customized market research data (e.g., by following the manner of customers’ engagement and how customers react to promotions).

Of all the consumer-related IoT initiatives, three types are most well-known: smart homes and appliances (Section 13.7), smart cities (Section 13.8), and autonomous vehicles (Section 13.9). For more on IoT and customers, see Miller (2018).

Chapter 13 • The Internet of Things as a Platform for Intelligent Applications 703

u SECTION 13.6 REVIEW QUESTIONS

1. Describe several enterprise applications. 2. Describe several marketing and sales applications. 3. Describe several customer service applications.

13.7 SMART HOMES AND APPLIANCES

The concept of the smart home has been in the limelight for several years, even before the concept of the IoT took a front stage.

A smart home is a home with automated components that are interconnected (frequently wirelessly), such as appliances, security, lights, and entertainment, and are centrally controlled and able to communicate with each other. For a description, see techterms.com/definition/smart_home.

Smart homes are designed to provide their dwellers with comfort, security, low energy cost, and convenience. They can communicate via smartphones or the Internet. The control can be in real time or at any desired intervals. Most existing homes are not yet smart, but they can easily and inexpensively be equipped to for at least partial smartness. Several pro- tocols enable connections; well-known ones are XIO, UPB, Z-Wave, and EnOcean. These products offer scalability, so more devices can be connected to the smart home over time.

For an overview, see techterms.com/definition/smart_home, smarthomeen- ergy.co.uk/what-smart-home, and Pitsker (2017).

In the United States followed by other countries, thousands of homes are already equipped with such systems.

Typical Components of Smart Homes

The following are typical components in smart homes:

• Lighting. Users can manage their home lighting from wherever they are. • TV. This is the most popular component. • Energy management. Home heating and cooling systems can be fully auto-

mated and controlled via a smart thermostat (e.g., see Nestnest.com/works- with-nest about its product Nest Learning Thermostat).

• Water control. WaterCop (watercop.com) is a system that reduces water dam- age by monitoring water leaks via a sensor. The system sends a signal to a valve, causing it to close.

• Smart speaker and chatbots (see Chapter 12). Most popular are Echo and Alexa, and Google Assistant.

• Home entertainment. Audio and video equipment can be programmed to respond to a remote control device. For instance, a Wi-Fi–based remote control for a stereo system located in a family room can command the system to play on speakers installed anywhere else in the house. All home automation devices per- form from one remote site and one button.

• Alarm clock. This tells kids to go back to sleep or to wake up. • Vacuum cleaner. Examples are iRobot Roomba, and LG Roboking vacuum; see

Chapter 2). • Camera. This allows residents to see what is going on in their homes anytime

from anywhere. Nest Cam Indoor is a popular product. Some smart cameras can even know how residents feel. See tomsguide.com/us/hubble-hugo-smart- home-camera,news-24240.html.

704 Part IV • Robotics, Social Networks, AI and IoT

The major components of smart homes are illustrated in Figure 13.4. Note that only a few homes have all of these components. Most common are home

security, entertainment, and energy management.

Example: iHealthHome

Security measures are common in assisted living facilities in senior communities and for seniors who live independently. For example, the iHealthHome Touch screen system col- lects data and communicates with caregivers using the company’s software. The system provides caregivers and physicians remote access to a person’s health data. Using this technology, the iHealthHome program also reminds seniors of daily appointments and when to take their medicine. The system also reminds people when to self-measure their blood pressure and how to stay in touch with their caregivers.

Smart Appliances

A smart appliance includes features that can remotely control the appliance operations, based on the user preferences. A smart appliance may utilize a Home Network or the Internet to communicate with other devices in the smart home.

• Refrigerator. An example of this is Instaview from LG, which is powered by Alexa. • Home security and safety. Such systems can be programmed to alert owners

to security-related events on their property. As noted, some security can be sup- ported by cameras for remote viewing of property in real time. Sensors can be used at home to detect intruders, keep an eye on working appliances, and perform several additional activities.

Lighting Energy

management Air-conditioning

Blinds

Ventilation

Heating

Motion sensor

IP Cam

Fire detector

Gas detector

Multimedia room

Leak detector

Sauna

GardenLaundry

Garage

Access control

FIGURE 13.4 The Components of a Smart Home.

Chapter 13 • The Internet of Things as a Platform for Intelligent Applications 705

McGrath (2016) provides an overview of smart appliances that includes all appli- ances from Haier (a large China-based manufacturer). Its goal is to make everything in a house communicate across other device makers. Examples are smart refrigerators, air conditioners, and washing machines. Haier offers a control board for all appliances re- gardless of their manufacturers. Apple is working on a single control for all smart appli- ances in a home.

GOOGLE’S NEST A leading manufacturer of IoT smart home applications is Google’s  Nest. The company is a producer of programmable self-learning, sensor- driven, Wi-Fi– enabled products. In the spring of 2018, the company had three major products:

• Learning thermostat. This device learns what temperature and humidity level that people like and controls the air conditioner/heating system accordingly. Google claims that its products provide an average energy savings of 13 per- cent, which could pay for the device in two years; see nest.com/thermostats/ nest-learning-thermostat/overview/?alt=3.

• Smoke detector and alarm. This device, which is controlled from a smartphone, tests itself automatically and lasts for about a decade. For details, see nest.com/ smoke-co-alarm/overview/.

• Nest.com. This Webcam-based system allows users to see what is going on in their homes from any location via smartphone or any desktop computer. The sys- tem turns itself on automatically when nobody is at home. It can monitor pets, babies, and so on. A photo recorder allows users to go back in time. For details, see nest.com/cameras/nest-cam-indoor/overview/. For how Nest can use a phone to find out when individuals leave home, see Kastrenakes (2016). For more on Nest, see en.wikipedia.org/wiki/Nest_Labs.

Examples of Available Kits for Smart Homes

Two popular smart-home starter kits are (Pitsker, 2017):

1. Amazon Echo. This includes Amazon Echo, Belkin Wemo Mini, Philips Hue white starter kit, Ecobee Lite, and Amazon Fire TV stick with Alexa voice remote. Total cost on October 2017 was $495.

2. Google Home. This includes Google Home, Smart Speaker, Belkin Wemo Mini, Philips Hue white starter kit, Nest learning thermostat, and Google Chromecast (for entertainment). Total cost on October 2017 was $520.

HOME APPLIANCES IN CONSUMER ELECTRONIC SHOW (CES) 2016–2018 The following smart appliances, some of which were exhibited at the CES show in Las Vegas in January 2016 (Morris 2016), 2017, and 2018, are:

• Samsung Smart fridge. Cameras check content; sensors check temperature and humidity.

• Gourmet robotic cooker. It does interesting cooking. • 10 in 1 device for the kitchen. This stirs food such as scrambled eggs and has

10 cooking styles (e.g., baking, sauce making). • LG HUM-BOT Turbo+. This can focus on an area in the home that needs special

attention. A camera monitors the home remotely while the owner is away (similar to Google’s Nest).

706 Part IV • Robotics, Social Networks, AI and IoT

• Haier R3D2 Refrigerator. According to Morris (2016), this refrigeration is not  the most practical one, but it has much of entertainment value. It looks like R3D2 in Star Wars. It can serve you a drink as well as provide lights and sounds.

• Instaview Refrigerator from LG. Powered by Alexa (enabled by voice), this includes a 29-inch LCD touch screen display. It provides functions such as deter- mining the expiration dates of food and notifying the user. For details, see Diaz (2017).

• Whirlpool’s smart top load washer. This fully automated machine has smart controls. It saves energy and even encourages philanthropy by sending a small amount of money to “Habitat for Humanity” each time washer is loaded.

• LG LDT8786ST dishwasher. This machine has camera whose sensors keep track of what has already been cleaned in order to save water. In addition, it pro- vides flexibility in operations.

The following are smart home trends:

• TVs that can be used as a smart Hub for home appliances is coming from Samsung.

• Dolby Atmos products include speakers, receivers, and other entertainment items. • DIY home smart security cameras make sure there is an intruder, not just the cat,

before alerting the police. • Water controls for faucets, sprinklers, and flood detectors are available. In addition,

a robot can teach users how to save water indoors (hydrao.com/us/en/).

For more about home automation, see smarthome.com/sh-learning-center-what-can- i-control.html. Various apps used for home control can be found at smarthome.com/ android_apps.html.

Smart components for the home are available at home improvement stores (e.g., Lowes) and can be purchased directly from manufacturers (e.g., Nest).

To facilitate the creation of smart components for the home, Amazon and Intel Corp. partnered in 2017 to provide developers with platforms to advance the smart home eco- system. For details, see pcmag.com/news/350055/amazon-intel-partner-to-advance- smart-home-tech/.

For smart appliances at CES 2018, watch the video at youtube.com/watch? v=NX-9LivJh0/.

A Smart Home Is Where the Bot Is

The virtual personal assistant that we introduced in Chapter 12 enables people to con- verse by voice with chatbots such as Alexa/Echo and Google Assistant. Such assistants can be used to manage appliances in smart homes.

In a comprehensive smart home, devices not only meet household needs but also are able to anticipate them. It is predicted that in the near future, an AI-based smart home will feature an intelligent and coordinated ecosystem of bots that will manage  and perform household tasks and may even be emotionally connected with people. For a prediction of the future bots, see Coumau et al. (2017). Amazon and Intel joined forces to develop such smart home ecosystems that include NLP capabilities.

Smart homes will also have smart robots that can serve people snacks, help take care of people who are handicapped, and even teach children different skills.

Chapter 13 • The Internet of Things as a Platform for Intelligent Applications 707

• Compatibility. There are too many products and vendors to choose from, mak- ing potential buyers confused. Many of these products do not “speak” to each other, so more industry standards are needed. In addition, it is difficult to match the products with consumers’ needs.

• Communication. Different consumers have different ideas on what the smart home should be. Therefore, the capabilities and benefits of a smart home need to be clearly communicated to users.

• Concentration. Brands need to concentrate on population segments that are most interested in smart homes (e.g., Gen Y).

Barriers to Smart Home Adoption

The potential of smart homes is attractive, but it will take some time before there will be many of them. The following are some limiting barriers, per Vankatakrishnan (2017).

In addition are the issues of cost justification, invasion of privacy, security, and ease of use. For the future of smart homes, including the role of Amazon and Walmart, and how the smart home will shop for itself, see Weinreich (2018).

Smart homes, appliances, and buildings can be featured in smart cities, the subject of our next section.

u SECTION 13.7 REVIEW QUESTIONS

1. Describe a smart home. 2. What are the benefits of a smart home? 3. List the major smart appliances. 4. Describe how Nest works. 5. Describe the role of bots in smart homes.

13.8 SMART CITIES AND FACTORIES

The idea of smart cities took off around 2007 when IBM launched its Smart Planet project and Cisco began its Smart Cities and Communities program. The idea is that in smart cities, digital technologies (mostly mobile based) facilitate better public services for citizens, better utilization of resources, and less negative environmen- tal impact. For resources, see ec.europa.eu/digital-agenda/en/about-smart-cities. Townsend (2013) provides a broad historical look and coverage of the technologies. In an overview of his book, he provides the following examples: “In Zaragoza, Spain, a ‘citizen card’ can get you on the free city-wide Wi-Fi network, unlock a bike share, check a book out of the library, and pay for your bus ride home. In New York, a guerrilla group of citizen-scientists installed sensors in local sewers to alert you when storm water runoff overwhelms the system, dumping waste into local waterways.” According to a prediction made by Editors (2015), smart cities would use 1.6 billion connected things in 2016. Finally, smart cities can have several smart entities such as universities and factories (see Lacey, 2016). For more on smart cities, see Schwartz (2015). In addition, watch the video “Cisco Bets Big on ‘Smart Cities’” at money.cnn. com/video/technology/2016/03/21/cisco-ceo-smart-cities.cnnmoney. Another video to watch is “Smart Cities of the Future” (3:56 min.) at youtube.com/ watch?v=mQR8hxMP6SY. A more detailed video on San Diego (44:06 minutes) is at youtube.com/watch?v=LAjznAJe5uQ.

708 Part IV • Robotics, Social Networks, AI and IoT

In over seven years, the city of Amsterdam (The Netherlands) was transformed into a smart city using information technologies. This case describes the steps the city took from 2009 to 2016 to become a smart city, as reported by MIT Sloan School of Management. The city initiative included projects in the following catego- ries: mobility, quality of living, transportation, security, health, and economy as well as infrastructure, big and open source data, and experimental living labs.

The major findings of the MIT team regarding Amsterdam’s transformation were:

• Private-sector data are critical for changing policy. The major categories of the project in- volved nongovernmental entities (e.g., using a GPS provider to manage traffic). For example, the private sector was involved in a project to change traffic situations (reduction of 25 percent in the number of cars and an increase of 100 percent in the number of scooters, in five years). • It is necessary to have chief technology

officers in smart cities. Smart cities re- quire the collection of large amounts of data using several tools and algorithms. Issues such as cost and security are critical.

• Expectations of the contribution of the IoT, Big Data, and AI, need to be man- aged. Citizens expect rapid changes and improvement in areas ranging from parking to traffic. Data collection is slow, and chang- es are difficult to implement.

• Smart city initiatives must start with data inventory. The problem in Amster- dam was that data were stored in 12,000 databases across 32 departments. These were organized differently on different hardware, so data inventory was needed. This initial activity was boring and tedious and had no immediate visible payoff.

• Pilot projects are an excellent strategy. Pilot projects provide lessons for future

projects. The city had over 80 pilot proj- ects, for example, collecting different types of trash and placing them in different colored bags. Successful projects are scaled up in size.

• Citizen input is a critical success factor. There are several ways to encour- age citizens to provide input. Involvement of universities and research institutions is also critical. In addition, social media net- works can be used to facilitate citizens’ en- gagement.

The smart city initiative may be only in its beginning, but it is already improving the quality of life of residents and increasing the economic growth of the city. A critical success factor of the initiative was the willingness of the city officials to share their data with technology companies.

IoT was a major component in the projects. First, it enabled the flow of data from sensors and databases for analytic processing. Second, IoT enables autonomous vehicles of all kinds, which contribute to the reduction of pollution, vehicle acci- dents, and traffic jams. Finally, IoT provides real-time data that help decision makers develop and improve policies. In April 2016, the city won Europe’s “Capital of Innovation” award (a prize of 950,000 euros).

Sources: Compiled from Brokaw (2016), Fitzgerald (2016), amsterdamsmartcity.com, and facebook.com/amsterdam smartcity.

Questions for Case 13.3

1. Watch the video at youtube.com/watch?v= FinLi65Xtik/ and comment on the technologies used.

2. Get a copy of the MIT case study at sloanreview. m i t . e d u / c a s e - s t u d y / d a t a - d r i ve n - c i t y - management/. List the steps in the process and the applications that were likely used in IoT.

3. Identify the smart components used in this project.

Application Case 13.3 Amsterdam on the Road to Become a Smart City

Cities cannot become smart overnight, as illustrated in Application Case 13.3, which presents the case of Amsterdam and its evolution into a smart city.

In many countries, governments and others (e.g., Google) are developing smart city appli- cations. For example, India has begun to develop 100 smart cities (see enterpriseinnovation. net/article/india-eyes-development-100-smart-cities-1301232910).

Chapter 13 • The Internet of Things as a Platform for Intelligent Applications 709

Smart Buildings: From Automated to Cognitive Buildings

IBM’S COGNITIVE BUILDINGS In a white paper (IBM, 2016), IBM discussed the use of IoT to make cognitive buildings, which are able to learn the behavior of a building’s system in order to optimize it. The cognitive building does so by autonomously integrating the IoT devices with the IoT operation. Such integration enables the creation of new business pro- cesses and increases the productivity of existing systems. Based on the concept of cognitive computing (Chapter 6), IBM describes the maturity of the technology as a continuation of the phase that started with automated buildings (1980 to 2000), the creation of smart build- ing (2000 to 2015), and finally, cognitive building (beginning in 2015). The process is illus- trated in Figure 13.5. The figure also shows the increased capabilities of buildings over time.

The highlights of a cognitive building are:

• By applying advance analytics, buildings can provide insights in near real time. • It learns and reasons from data and interacts with humans. The system can detect

and diagnose abnormal situations and propose remedies. • It has the ability to change building temperature subject to humans’ preferences. • It is aware of its status and that of its users. • It is aware of its energy status and adjusts it to be comfortable to dwellers. • Its users can interact with the building via text messages and voice chatting. • Robots and drones are starting to operate inside and outside the building without

human intervention.

A major collaborator of IBM is Siemens (from Germany). The companies concentrate on global issues related to the use of IoT to enhance building performance.

Smart Components in Smart Cities and Smart Factories

The major objective of smart cities is to automate as many as possible public services such as transportation, utilities, social services, security, medical care, education, and economy. So, in the smart city overall project one may find several subprojects, some of which are independent of the master project.

Example

Hong Kong has a project called a smart mobility for the improvement of road safety. A consortium of private and public organizations has introduced Intelligent Transport

Learn Behavior Predictive control down to desk level Understand energy flow and building occupancy Consider comfort preferences of users Collect context such as weather and meetings Too many data points even for advanced analytics

Analyze Energy Consumers Understand consumption of rooms and central assets Only primary data points are analyzed

Visualize KPI Good for ratings Allows identifying general issues

Automated Buildings (1980–2000)

Smart Buildings (2000–2015)

Cognitive Buildings (. 2015)

Bad for identifying energy waste

FIGURE 13.5 IBM’s Cognitive Building Maturity Framework. Source: IBM. “Embracing the Internet of Things in the new era of cognitive buildings.” IBM Global Business Services, White Paper, 2016. Courtesy of International

Business Machines Corporation, © International Business Machines Corporation.Used with permission.

710 Part IV • Robotics, Social Networks, AI and IoT

Services, including a warning mechanism for collision, and control assistance for finding parking. The system also manages speed and lane violations and traffic congestion. All of these increase safety and efficiency. For details, see Estopace (2017b).

Transportation is a major area in which analytics and AI can make cities smarter. Other areas include economic development, crime fighting, and healthcare. For details, see SAS (2017).

Other examples of smart city components can be found in a smart university, smart medical centers, smart power grid, and in airports, factories, ports, sport arenas, and smart factories. Each of these components can be treated as an independent IoT project, and/or as a part of the smart city overall project.

SMART (DIGITAL) FACTORIES Automation of manufacturing has been with us for gen- erations. Robots are making thousands of products from cars to cellphones. Tens of thousands of robots can be found in Amazon’s distribution centers. Therefore, it is not surprising that factories are getting smarter with AI technologies and IoT applications. As such they may be considered a component of smart cities and may be interrelated with other components, such as clean air and transportation.

A smart factory, according to Deloitte University Press, is “a flexible system that can self-optimize performance across a broader network, self-adapt to and learn from new conditions in real or near real time, and autonomously run entire production pro- cesses.” For details, see the free Deloitte e-book at DUP_The-smart-factory.pdf. For a primer, see https://www2.deloitte.com/insights/us/en/focus/internet-of-things/ technical-primer.html.

Tomás (2016) provides a vision of what industrial production will look like in the future. It will be essentially fully digitized and connected, fast, and flexible. The major idea is that there will be a command center in a factory equipped with AI technologies. The AI, combined with IoT sensors and information flow, will enable optimal organization and sequencing of business processes. The entire production chain, from raw material suppli- ers, logistics, and manufacturing to sales, will be connected to IoT systems for planning, coordination, and control. Planning will be based on analytic predictions of demand.

Production processes will be automated as much as possible and wirelessly controlled. Logistics will be provided on demand quickly, and quality control will be automated. IoT combined with sensors will be used for both predictive and preventive maintenance. Some of these elements exist in advanced factories, and more factories will be smarter in the future.

For more on smart factories, see Libelium (2015) and Pujari (2017). For the smart factory of the future, read belden.com/blog/industrial-ethernet/topic/smart- factory-of-the-future/page/0.

The use of IoT in the factory is illustrated in the video “Smart Factory Towards a Factory of Things” at youtube.com/watch?v=EUnnKAFcpuE (9:10 min.).

Smart factories will have different business processes, new technology solutions, different people-machine interactions, and a modified culture. For the transformation process to a smart factory, see Bhapkar and Dias (2017). The accounting firm Deloitte (dupress.deloitte.com/smart-factory) provides a diagram that illustrates “the major characteristics of a smart factory” (Figure 13.6).

Example: Smart Bike Production in a Smart Factory

The world demand for smart bikes is increasing rapidly, especially in smart cities. Mobike is the world’s first and largest bike-sharing company. To meet the demand, the com- pany is working with Foxconn Technology Group to make the bike production smarter. The smart manufacturing involves the creation of a global supply chain from raw ma- terials to production to sales. Foxconn is known for its high-technology expertise in

Chapter 13 • The Internet of Things as a Platform for Intelligent Applications 711

providing efficient manufacturing processes in a cost-efficiency production. It optimizes Internet-driven smart manufacturing. The production output is expected to double in the near future. For details, see Hamblen (2016) and enterpriseinnovation.net/article/ foxconn-drives-mobike-smart-bike-production-1513651539.

EXAMPLES OF SMART CITY INITIATIVES Smart city initiatives are diversified, as explained earlier. For examples, see Application Case 13.4.

CONNECTED Continuously pull traditional datasets along with new sensor and location-based datasets Real-time data-enabling collaboration with suppliers and customers Collaboration across departments (e.g., feedback from production to product development)

Reliable, predictable production capacity Increased asset uptime and production efficiency Highly automated production and material handling with minimal human interaction Minimized cost of quality and production

Live metrics and tools to support quick and consistent decision making Real-time linkages to customer demand forecasts Transparent customer order tracking

Predictive anomaly identification and resolution Automated restocking and replenishment Early identification of supplier quality issues Real-time safety monitoring

Flexible and adaptable scheduling and changeovers Implementation of product changes to see impact in real time Configurable factory layouts and equipment

OPTIMIZED

TRANSPARENT

PROACTIVE

AGILE

•

•

•

• •

•

•

•

•

•

•

• •

•

•

•

•

FIGURE 13.6 Five Key Characteristics of a Smart Factory (Deloitte). Source: Burke, Hartigan, Laaper, Martin, Mussomeli, Sniderman, “The smart factory: Responsive, adaptive, connected

manufacturing,” Deloitte Insights (2017), https://www.deloitte.com/insights/us/en/focus/

industry-4-0/smart-factory-connected-manufacturing.html. Used with permission.

IBM has been supporting smart city initiatives for several years. The following examples are com- piled from Taft’s slide show (eweek.com/cloud/ how-ibm-is-making-cities-smarter-worldwide).

• Minneapolis (United States). The initiative supports more effective decisions for the city’s

resource allocation. In addition, it aligns the operations of multiple departments working on the same project. IBM is providing AI-based pattern recognition algorithms for problem solving and performance improvement.

• Montpellier (France). IBM’s software is help- ing the city in its initiatives of water manage-

Application Case 13.4 How IBM Is Making Cities Smarter Worldwide

(Continued )

712 Part IV • Robotics, Social Networks, AI and IoT

A major area of improvement in a smart city is transportation.

Improving Transportation in the Smart City

A major problem in many cities is the increased number of vehicles and the inability to ac- commodate all of them effectively. Building more roads could add more pollution and lead to traffic jams. Public transportation can help alleviate the problem but may take years to com- plete. Quick solutions are needed. In the opening case to Chapter 2, we introduced Inrix. The Inrix company uses AI and other tools to solve transportation problems. It collects data from stationary sensors along roads and from other sources. In some smart cities, innovators have already placed air quality sensors on bicycles and cars. Sensors also are taking data from cars on the roads to help generate data that can analyzed and results are transmitted to drivers. An example of another innovative project is provided in the following examples.

Example 1

Valerann, an Israeli start-up, developed smart road studs to replace the reflective studs of today’s technology. Smart studs can transmit information of what they sense about what

ment, mobility (transportation), and risk manage- ment (decision making). The rapidly growing city must meet the increasing demand for services. To do this efficiently, IBM provides data analysis and interpretation of activities, research institutions, and other partners in the region.

• Stockholm (Sweden). To reduce traffic prob- lems, IBM technologies are optimally matching demands and supplies. The initiative uses sen- sors and IoT to alleviate the congestion problem.

• Dubuque (United States). Several initia- tives were conducted for efficient use of re- sources (e.g., utilities) and management of transportation problems.

• Cambridge (Canada). The city is using IBM’s “Intelligent Infrastructure Planning” for conducting business analytics and decision sup- port technologies. Using AI-based algorithms, the city can make better decisions (e.g., repair or replace assets). In addition, IBM smart tech- nologies help to improve project coordination.

• Lyon (France). Transportation management is a major project in any big city and a target for most smart city initiatives. Smart technolo- gies provide transportation staff with effective real-time decision support tools. This helped reduce traffic congestion. Using predictive ana- lytics, future problems can be forecasted, so, if they occur, they can be solved quickly.

• Rio de Janeiro (Brazil). To manage and coordinate the operations of 30 city depart- ments is a complex undertaking. IBM tech- nologies support a central command center for the city that plans operations and handles emergencies in all areas.

• Madrid (Spain). To manage all its emergen- cy situations (fire, police transportation, hospi- tals), the city created a central response center. Data are collected by sensors, GPS, surveillance cameras, and so on. The center was created af- ter Madrid’s 2004 terrorist attack and is managed with the support of IBM smart technologies.

• Rochester (United States). The city police department is using IoT and predictive analy- sis to forecast when and where crimes is like- ly to be committed. This AI-based system has proven to be accurate in several other cities.

These examples illustrate the utilization of IBM’s Smarter Cities framework in several areas by smart city initiatives. Note that IBM Watson is using IoT for many of its own projects.

Questions for Case 13.4

1. List the various services that are improved by IoT in a smart city.

2. How do the technologies support decision making?

3. Comment on the global nature of the examples.

Application Case 13.4 (Continued)

Chapter 13 • The Internet of Things as a Platform for Intelligent Applications 713

is occurring on the roads. Eventually, the studs will be incorporated with autonomous ve- hicles. The smart studs cost more than reflective studs but have a longer life. For details, see Solomon (2017).

Example 2

Smart Mobility Consortium (Hong Kong) works on mobility in the smart city of Hong Kong. More than 10 million people there use the public and private transportation sys- tems every day. This transportation project includes several smart subsystems for parking, collision warning, and alerts for speeders and lane changing violators. For details, see Estopace (2017b).

Combining Analytics and IoT in Smart City Initiatives

Like in many IoT initiatives, it is necessary to combine analytics and IoT. A notable example is IBM Watson. Another one is the SAS platform.

Example: The SAS Analytics Model for Smart Cities

The amount of data collected by IoT networks in cities can be enormous. Data are col- lected from many sensors, computer files, people, databases, and so on. To make sense of these data, it is necessary to use analytics, including AI algorithms. SAS is using a seven-step process divided into three major phases: Sense, Understand, and Act. The fol- lowing are definitions of these (condensed from SAS, 2017).

• Sense. Using sensors, sense anything that matters. SAS analyzes the collected data. The data go through intelligent filters for cleanliness so that only relevant data go to the next phase. IoT collects and transfers the data from the sensors.

• Understand the signals in the data. Using data mining algorithms, the entire rel- evant ecosystem is analyzed for pattern recognition. The process can be complex as the data collected by IoT sensors are combined with data from other sources.

• Act. Decisions can be made quickly as all relevant data are in place. SAS decision management tools can support the process. Decisions range from alerts to auto- mated actions.

The SAS process is illustrated in Figure 13.7. For more on analytics and IoT combination, see SAS Analytics for IoT at https://www.sas.com/en_us/insights/big-data/internet- of-things.html. For additional information, see Henderson (2017).

Bill Gates’ Futuristic Smart City

In November 2017, Bill Gates purchased 60,000 acres of land west of Phoenix, Arizona, where he plans to construct a futuristic city from scratch. The city will be a model and place for research.

Technology Support for Smart Cities

A large number of vendors, research institutions, and governments are providing technol- ogy support for smart cities. Here are few examples.

TECHNOLOGY SUPPORT BY BOSCH CORP. AND OTHERS Bosch Corp (of Germany), a major supplier of automotive parts, presented several innovations related to smart cities at CES 2018.

According to Editors (2018), revenues of global smart cities with IoT technology will exceed $60 billion by 2026.

714 Part IV • Robotics, Social Networks, AI and IoT

Finally, in smart cities, connected and self-driven vehicles will be everywhere (see Hamblen, 2016 and the next section).

u SECTION 13.8 REVIEW QUESTIONS

1. Describe smart city. 2. List some benefits of a smart city to the residents. 3. What is the role of IoT in smart city initiatives? 4. How are analytics combined with IoT? Why? 5. Describe smart and cognitive buildings. 6. What is a smart factory? 7. Describe technology support to smart cities.

13.9 AUTONOMOUS (SELF-DRIVING) VEHICLES

Autonomous vehicles, also known as driverless cars, robot-driven cars, self-driving cars, and autonomous cars, are already on the roads in several places. The first commer- cial autonomous car project was initiated by Google (named Google Chauffeur) and is becoming a reality, with several U.S. states preparing to allow them on the road. France, Singapore, China, and several other countries already have these cars and buses on their roads. These cars are electric, and they can create a revolution by reducing emissions, accidents, fatalities (an estimate of about 30,000 fatalities a year, worldwide), and traf- fic jams (e.g., see Tokuoka, 2016). Thus far, these cars are being tested in several cities worldwide and in some cities are already on the roads. Experts estimate that 10 million such cars will be on the roads in the United States by 2020, and China is planning for 30 million cars by 2021.

The Developments of Smart Vehicles

The initial efforts to commercialize a self-driving car were started by Google in the 1990s. These efforts can be seen today in Waymo’s story in Application Case 13.5.

Data Access

Intelligent Filter/

Transformation

Streaming Model

Execution

Data Storage

Ad Hoc Analysis

Model Development/ Deployment

Alerts/ Reporting

Sense Understand Act

FIGURE 13.7 SAS Supports the Full IoT Analytics Life Cycle for Smart Cities (SAS). Source: Courtesy of SAS Institute Inc. Used with permisison.

Chapter 13 • The Internet of Things as a Platform for Intelligent Applications 715

Waymo is a unit of Alphabet (previously called Google) that is fully dedicated to the Google self- driving car project. Almost 20 years ago, Google, with the help of Stanford University, started to work on this project. The idea received a boost in 2005 when DARPA awarded its Grand Challenge prize to the project. Then, the U.S. Department of Defense awarded it a $2 million prize. Google pio- neered physical experiments in 2009 after conduct- ing computer simulation for several years when it ran self-driving cars 2.5 billion virtual miles. The next step was to get legislation to allow autono- mous vehicles on the roads. By 2018, 10 states had passed such laws. Some allow robot-driven cars only in certain areas. Self-driving cars (see a Waymo car in Figure 13.8) with robot-only chauf- feurs were tested in early 2018 by Waymo in the Phoenix, Arizona, area. First, corporate engineering will be in the driver’s seat; but, around November 2018, the cars were expected to be completely driverless. The company was ready to start running commercial minivans in five states in 2018. By the end of 2018, Waymo vans were expected to pick up regular passengers who volunteered to take the service (called Early Rider Program), although most travelers are still skeptic.

This works in the following way. Company technicians, acting like regular riders, order service via a mobile app. The AI mechanism figures out how

the vehicle will get to the requested caller as well as how it will self-drive to the requested destination.

Waymo, the pioneer of autonomous vehicles, collaborated with Chrysler (using Chrysler Pacifica minivans). The computing power is provided by Intel (with its Mobileye division). The high cost of the cars will limit their use initially to commercial uses. However, Waymo already has agreed to manage Avis’s fleet of self-driving minivans. Also, realizing the power of ride-sharing services, Waymo is work- ing with Lyft on new autonomous vehicles. Finally, Waymo is partnering with AutoNation to provide maintenance and road services for Waymo cars.

Note: On the legal dispute involving Uber, see the opening case of Chapter 14.

Sources: Compiled from Hawkins (2017), Ohnsman (2017), and Khoury (2018).

Questions for Case 13.5

1. Why did Waymo first use simulation?

2. Why was legislation needed?

3. What is the Early Rider Program?

4. Why will it take years before regular car owners will be able to enjoy a ride in the back seat of their self-driving cars?

5. Why are Lyft, Uber, and Avis interested in self- driving cars?

Application Case 13.5 Waymo and Autonomous Vehicles

FIGURE 13.8 Waymo (Google) Self-Driving Car. Source: SiliconValleyStock/Alamy Stock Photo.

716 Part IV • Robotics, Social Networks, AI and IoT

TECHNOLOGY INSIGHTS 13.2 Toyota and Nvidia Corp. Plan to Bring Autonomous Driving to the Masses

It is not surprising that Toyota is interested in smart cars. As a matter of fact, the company’s cars are expected to be on the market in 2020. Toyota plans to produce several types of autonomous vehicles. One type will be for elderly and disabled people. Another type will have the ability to drive completely autonomously or be an assistant (with a mechanism called “guardian angel”) to drivers. For example, it will have the ability to take full control when the driver falls asleep, or when it senses that an accident is coming. A tired driver will be able to use Alexa (or a similar device) to tell the guardian angel to take over.

Autonomous vehicles need a smart control system, and this is where Nvidia enters the pic- ture. Autonomous cars need to process a vast amount of data collected by sensors and cameras in real time. Nvidia pioneered a special AI-based supercomputer (called Drive PX2) for this pur- pose. The computer includes a special processor (called Xavier) that can power the autonomous driving gear of the cars. The partnership with Toyota enables Nvidia to leverage the power of its processor to apply AI to the autonomous cars.

Nvidia’s supercomputer has an AI algorithm-based special operating system that includes a cloud-based 3D map with high definition. With these capabilities, the car’s “brain” can com- prehend its driving surroundings. Since a car can also exactly identify its own location, it will know about any potential hazard (e.g., road work or a vehicle coming toward it). The operating system is being constantly updated, so it makes the car smarter (AI learning capability).

The Xavier system provides the car’s “brain” on a special chip (called Volta), which can deliver 30 trillion deep learning operations per second. Thus, it can process complex AI algorithms involv- ing machine learning. Nvidia is expected to use Volta to open a new, powerful era in AI computing.

Source: Compiled from Korosec (2017) and blogs.nvidia.com/blog/2016/09/28/Xavier/.

Questions for DisCussion

1. What does a car need to have in order to be autonomous?

2. What is the contribution of Nvidia to self-driving cars?

3. What is the role of Xavier?

4. Why does the process use a supercomputer?

Despite the required complex technology, several car manufacturers are ready to sell or operate such cars soon (e.g., BMW, Mercedes, Ford, GM, Tesla, and of course—Google).

Developments related to driverless vehicles follow:

• Uber and other ride-sharing companies plan for self-driving cars. • Mail is delivered to homes by self-driving cars; see uspsoig.gov/blog/

no-driver-needed. • Driverless buses are being tested in France and Finland. Watch money.cnn.com/

video/technology/2016/08/18/self-driving-buses-hit-the-road-in-helsinki. cnnmoney about self-driving buses in Helsinki.

• Self-driving taxis already operate in Singapore.

The Self Drive Act is the first national law in the United States pertaining to self- driving cars. It aims to regulate the safety of the passengers in autonomous vehicles. It opens the door for the production of 100,000 cars per year by 2021.

An example of how Nvidia works with Toyota’s initiative is presented in Technology Insights 13.2.

Chapter 13 • The Internet of Things as a Platform for Intelligent Applications 717

Flying Cars

While autonomous vehicles on the road may have considerable difficulties, there is re- search on flying cars. As a matter of fact, drones that can carry people already exist. As long as there is not much traffic in the air, there will be no traffic problem. However, the navigation of a large number of flying cars may be a problem. Airbus created a flying taxi demo in 2016 and Uber developed the concept and summarized it in a 98-page report released in October 2016. Toyota is also working on making a flying car. In January 2018, at the Las Vegas CES, Intel showed an autonomous passenger drone named Volocopter. This machine can be developed as an air taxi one day. For flying taxis in New Zealand, see Sorkin (2018).

Implementation Issues in Autonomous Vehicles

Autonomous vehicles such as cars, trucks, and buses are already on the roads in several cities worldwide. However, before we will see millions of them on the roads, it will be necessary to deal with several implementation issues. The following are reasons why full commercialization is going to take time:

• The cost of real-time 3D map technologies needs to be reduced and their quality needs to be increased.

• AI software must be nimble and its capabilities increased. For example, AI needs to deal with many unexpected conditions, including that of the behavior of drivers of other cars.

• Bray (2016) posted an interesting question: “Are customers, automakers and insur- ers really ready for self-driving cars?” Customers seem to acknowledge that such cars are coming. But they resist boarding one. However, some daring people expect these cars to do a better job than humans in driving.

• The technology needs more research, which is very expensive. One reason is that the many sensors in the cars and on the road need to be improved and their cost need to be reduced.

• The IoT is connecting many objects for autonomous vehicles, including those in clouds. The IoT systems themselves need to be improved. For example, data trans- mission delays must be eliminated. For more IT/AI generic implementation issues, see Chapter 14.

u SECTION 13.9 REVIEW QUESTIONS

1. What are self-driving vehicles? How are they related to the IoT? 2. What are the benefits of self-driving vehicles to drivers, society, and companies? 3. Why are Uber and similar companies interested in self-driving vehicles? 4. What AI technologies are needed to support autonomous vehicles? 5. What are flying cars? 6. List some implementation issues of autonomous vehicles.

13.10 IMPLEMENTING IoT AND MANAGERIAL CONSIDERATIONS

In this chapter, we presented a number of successful IoT-based applications. The results so far are more than encouraging, especially in areas such as monitoring equipment performance to improve its operation and maintenance (e.g., CNH in the opening vi- gnette and the IBM Watson case of elevators in Chapter 1). However, this is only the tip of the iceberg. As we indicated earlier, the IoT can change everything. In this section,

718 Part IV • Robotics, Social Networks, AI and IoT

we present some of the major issues that are related to successful IoT implementation. Although there is considerable excitement about the growth and the potential of the IoT, there are that managers should be aware of.

Major Implementation Issues

McKinsey’s Global Institute (Bughin et al., 2015) has put together a comprehensive Executive’s Guide to the Internet of Things. This guide identifies the following issues:

• Organizational alignment. Although it is true of several other technology initia- tives, with IoT, the opportunities for operational improvements and creating new business opportunities means that IT and operational personnel have to work as one team rather than separate functions. As noted by the guide’s authors, “IoT will challenge other notions of organizational responsibilities. Chief financial, marketing, and operating officers, as well as leaders of business units, will have to be receptive to linking up their systems.”

• Interoperability challenges. Interoperability is a huge detriment thus far in the growth of IoT applications. Few IoT devices connect seamlessly with each another. Second, there are many technological issues regarding connectivity. Many remote areas do not yet have proper Wi-Fi connection. Issues related to Big Data process- ing are also responsible for slow progress in IoT adoption. Companies are trying to reduce data at the sensor level so that only a minimal amount goes into clouds. Current infrastructure hardly supports the huge amount of data collected by IoT. A related problem is retrofitting sensors on devices to be able to gather and transmit data for analysis. In addition, it will take time for consumers to replace their analog objects with new IoT digital smart products. As an example, it is easier for people to replace mobile phones than a car, kitchen appliances, and other things that can benefit from having a sensor and being connected to IoT.

• Security. Security of data is an issue in general, but it is an even bigger one in the context of IoT. Each device that is connected to IoT becomes another entry point for malicious hackers to get into a large system or at the least operate or corrupt a specific device. There are stories of hackers being able to breach and control auto- mated functions of a car or to control a garage door opener remotely. Such issues require that any large-scale adoption of IoT involve security considerations from the very beginning.

Given that the Internet is not well secured, applying IoT networks requires special security measures, especially in the wireless sections of the networks. Perkins (2016) summarizes the situation as follows: “IoT creates a pervasive digital presence connecting organizations and society as a whole. New actors include data scientists; external integra- tors; and exposed endpoints. Security decision makers must embrace fundamental prin- ciples of risk and resilience to drive change.” For a free e-book about IoT, see McLellan (2017b).

Additional issues follow.

• Privacy. To ensure privacy, one needs a good security system plus a privacy pro- tection system and policy (see Chapter 14). Both may be difficult to construct in IoT networks due to the large size of the networks and the use of the less protected Internet. For advice from top security experts, see Hu (2016).

• Connection of the silos of data. There are millions of silos of data on the Internet and many of them need to be interconnected in specific IoT applications. This issue is known as the need for a “fabric” and connectivity. This can be a complex issue for applications that involve many different silos belonging to dif- ferent organizations. Connectivity is needed in machine to machine, people to

Chapter 13 • The Internet of Things as a Platform for Intelligent Applications 719

people, people to machines, and people to services and sensors. For a discussion, see Rainie and Anderson (2017) and machineshop.io/blog/the-fabric-of-the- internet-of-things. For how the connection is done at IBM Watson, see ibm.com/ Internet-of-things/iot-solutions/.

• Preparation of existing IT architectures and operating models for IoT can be a complex issue in many organizations. For a complete analysis and guide on this subject, see Deichmann et al. (2015). Integrating IoT into IT is critical for the data flow needed by the IoT and IoT-processed data to flow back to actions.

• Management. As in the introduction of any new technology, the support of top management is necessary. Bui (2016) recommends hiring a chief data officer in order to succeed in IoT due to the need to deal with silos of data described earlier. Using such a top manager can facilitate information sharing across all business func- tions, roles, and levels. Finally, it solves departmental struggles to own and control the IoT.

• Connected customers. There is evidence of an increased use of IoT in mar- keting and customer relationships. In addition, the IoT drives increased customer engagement. According to Park (2017), a successful deployment of IoT for custom- ers requires “connected customers.” The connection needs to be for data, decisions, outcomes, and staff related to any contacts relevant to the IoT and marketing. The Blue Hill research organization provides a free report on this issue (see Park). IoT enables a better connection with key clients and improves customer service. Of special considerations are hospitality, healthcare, and transportation organizations.

Finnaly, Chui et al. (2018) provided suggestions in a recent study on how to succeed in IoT implementation.

With so many implementation issues, an implementation strategy is necessary.

Strategy for Turning Industrial IoT into Competitive Advantage

IoT collects large amounts of data that can be used to improve external business activities (e.g., marketing) as well as internal operations. SAS (2017) proposed a strategy cycle that includes the following steps:

1. Specify the business goals. They should be set with perceived benefits and costs so the initiatives can be justified. This step involves a high level of planning and exami- nation of resources. Initial return on investment (ROI) analysis is advisable.

2. Express an analytic strategy. To support ROI and prepare a business case, it will be necessary to plan how Big Data will be analyzed. This involves the selection of an analytic platform, which is a critical success factor. An examination of emerging AI technologies, such as deep learning, may be conducted. An appropriate selection will ensure a powerful IoT solution.

3. Evaluate the needs for edge analytics. Edge analytics is a technology that is needed for some, but not all, applications. It is designed to introduce real-time capabilities to the applications. It also filters data to enable automated decision making, frequently in real time because only relevant data results from the filtering.

4. Select appropriate analytics solutions. There are numerous analytic solutions on the market offered by many vendors. In using one or several for IoT, it is necessary to consider several criteria such as fitness for IoT, ease of deployment, ability to minimize project risks, sophistication of the tools, and connection to existing IT systems (e.g., the quality of IoT gateways). Sometimes it is a good idea to look at a group of vendors that offer combined products (e.g., SAS and Intel). Finally, appropriate infrastructures, such as high-performance cloud servers and storage systems, need to be examined. These must work together as a scalable, effective, and efficient platform.

720 Part IV • Robotics, Social Networks, AI and IoT

5. Continues improvement closes the loop. Like in any strategy cycle, performance should be monitored, and improvements in various steps of the process need to be considered, especially since IoT is evolving and changing rapidly. The extent of goal achievement is an important criteria and upgrading the goals should be considered.

A summary of the process is provided in Figure 13.9. Weldon (2015) suggests the following steps for successful IoT implementation:

Management activities for using IoT for competitive

advantage

Specify business goals and objectives

Device analytical

strategy and plan

Activate continuous

improvements Measure

performance

Select appropriate analytical products

Evaluate the need for

edge analytics

FIGURE 13.9 The IoT Strategy Cycle.

• Develop a business case to justify the IoT project including a cost-benefit analysis and a comparison with other projects.

• Develop a working prototype. Experiment with it. Learn and improve it. • Install the IoT in one organizational unit; experiment with it. Learn lessons. • Plan an organization-wide deployment if the pilot is a success. Give special atten-

tion to data processing and dissemination.

The Future of the IoT

With the passage of time, we see an increasing number of IoT applications, both external and internal to organizations and enterprises. Because all IoT networks are connected to the Internet, it will be possible to have some of the networks connected to each other, creating larger IoTs. This will create growth and expansion opportunities for many organizations.

AI ENHANCEMENT OF IoT There are several areas of potential development. One area where AI will enhance IoT is in its ecosystem. Many IoT applications are complex and could be improved with machine learning that can provide insights about data. In ad- dition, AI can help in creating devices (“things”) that can self-diagnose problems and even repair them. For further discussion, see Martin (2017). Another future benefit of AI when combined with IoT is “shaping up to be a symbiotic pairing” (Hupfer, 2016). This pairing can create cognitive systems that are able to deal with and understand data that

Chapter 13 • The Internet of Things as a Platform for Intelligent Applications 721

conventional analytics cannot handle. The AI and IoT combination can create an embod- ied cognition that injects AI capabilities into objects (such as robots and manufacturing machines) to enable the objects to understand their environments and then self-learn and improve their operation. For details, see Hupfer (2017). Finally, AI can help the integra- tion of IoT with other IT systems.

A final word! By now you are probably interested to know about getting a job in IoT. Yes, there is a shortage of IoT experts, and annual salaries can range from $250,000 to $500,000. For 2017 data, see Violino (2017).

Chapter Highlights

• The IoT is a revolutionary technology that can change everything.

• The IoT refers to an ecosystem in which a large number of objects (such as people, sensors, and computers) are interconnected via the Internet (frequently wirelessly). By the years 2020 to 2025, there could be as many as 50 billion connected objects. Subsystems of such connected things can be used for many purposes.

• Use of the IoT can improve existing business pro- cesses and create new business applications.

• Billions of things will be connected to the Internet, forming the IoT ecosystem.

• Things on the IoT will be able to communicate, and the structure will enable a central control to manipulate things and support decision making in IoT applications.

• The IoT enables many applications in industry, services, and governments.

• IoT applications are based on analysis of data collected by sensors or other devices that flow over the Internet for processing.

• Sensors can collect data from a large number of things (e.g., over 1 million elevators in the open- ing case of Chapter 1).

• Major efforts are needed to connect the IoT with other IT systems.

• IoT applications can support decisions made by equipment manufacturers and by the users

of equipment. (See the opening vignette of this chapter.)

• IBM Watson is a major provider of IoT appli- cations in many industries and services (e.g., medical research). It was projected to reach over 1 billion users by the end of 2018.

• Smart appliances and homes are enabled by IoT.

• Smart city projects worldwide are supported by IoT, increasing the quality of life for res- idents of the cities and supporting the deci- sion making of city planners and technology providers.

• Self-driven cars may reduce accidents, pollution, traffic jams, and transportation costs. Self-driving cars are not fully implemented yet, but some were introduced in 2018.

• Smart homes and appliances are popular. For a small cost, owners can use several applications from home security to controlling appliances in their homes.

• The concept of smart cities is being developed globally with projects in countries such as India, Germany, and the United States and the city- state of Singapore. The objective of smart cit- ies is to provide a better life for their residents. Major areas covered are transportation, health- care, energy saving, education, and government services.

Key Terms

autonomous vehicles (driverless cars)

Internet of Things Internet of Things ecosystem

radio-frequency identification (RFID)

sensor smart appliance

smart cities smart factory smart homes smart sensors

722 Part IV • Robotics, Social Networks, AI and IoT

Questions for Discussion

1. Compare the IoT with regular Internet. 2. Discuss the potential impact of autonomous vehicles on

our lives. 3. Why must a truly smart home have a bot? 4. Why is the IoT considered a disruptive technology? 5. Research Apple Home Pod. How does it interact with

smart home devices? 6. Alexa is now connected to smart home devices such

as thermostats and microwaves. Find examples of other

appliances that are connected to Alexa and write a report.

7. Discuss the objective of smart cities to conserve the earth’s limited resources.

8. What are the major uses of IoT? 9. Accidents involving driverless cars slow down the

implementation of the technology. Yet, the technology can save hundreds of thousands of lives. Is the slow- down (usually driven by politicians) justifiable? Discuss.

Exercises

1. Go to theinternetofthings.eu and find information about the IoT Council. Write a summary of it.

2. Go to https://www.ptc.com/en/resource-center or other sources, and select three IoT implemented cases. Write a summary of each.

3. AT&T is active in smart city projects. Investigate their activities (solutions). Write a summary.

4. It is said that the IoT will enable new customer service and B2B interactions. Explain how.

5. The IoT has a growing impact on business and e-commerce. Find evidence. Also read Jamthe (2016).

6. Find information about Sophia, a robot from Hanson Robotics. Summarize her capabilities.

7. Examine the Ecobee thermostat and its integration with Alexa. What are the benefits of the integration? Write a report.

8. Enter smartcitiescouncil.com. Write a summary of the major concept found there; list the major enablers and the type of available resources.

9. Find the status of Bill Gates’s futuristic smart city. What are some of its specific plans?

10. City Brain is the name of Alibaba’s platform for smart cities. One project has been adopted in China and Malaysia. Find information and write a report.

11. Find the status of delivering pizza by self-driving cars. Check Domino’s Pizza news.

12. India has many IoT applications, including proj- ects for 100 smart cities. Read the 2016 status report atenterpriseinnovation.net/article/internet-things- next-big-wave-india-1270947471/ and find more recent information about it. Why do you think IoT is so widespread in India? Write a report.

13. Read the Blue Hill report (Park, 2017) and summarize all the issues related to IoT.

14. Find the status of smart cities as it is related to IoT and Cisco. Write a report.

15. Watch the video atyoutube.com/ watch?v=ZJr0X3XBMmA (14:36 min.). Write a sum- mary about the five smart devices.

16. Watch the video “Smart Manufacturing” (22 min.) at youtube.com/watch?v=SfVUkGoCA7s and summa- rize the lessons learned.

17. The competition for creating and using autonomous cars is intensifying globally. Find 12 companies that are competing in this field.

18. Enter McKinsey Global Institute mckinsey.com/mgi/ overview and find recent studies on IoT. Prepare the summary.

19. AT&T is trying to connect autonomous vehicles to smart cities. Find information on the progress of this project. Identify the benefits and the difficulties.

References

Ashton, K. How to Fly a Horse: The Secret History of Creation, Invention and Discovery. New York City, NY: Doubleday, January 2015.

Bhapkar, R., and J. Dias “How a Digital Factory Can Trans- form Company Culture.” McKinsey & Company, September 2017.

Bordo, M. “Israeli Air Force Works on Battlefield IoT Technol- ogy.” ReadWrite.com, June 21, 2016.

Bray, E. “Are Consumers, Automakers and Insurers Really for Self-Driving Cars?” Tech Crunch, August 10, 2016.

Brokaw, L. “Six Lessons from Amsterdam’s Smart City Initia- tive.” MIT Sloan Management Review, May 25, 2016.

Bughin, J., M. Chui, and J. Manyika. “An Executive’s Guide to the Internet of Things.” McKinsey Quarterly, August 2015.

Bui, T. “To Succeed in IoT, Hire a Chief Data Officer.” Tech Crunch, July 11, 2016.

Burkacky, O., et al. “Rethinking Car Software and Electronics Architecture.” McKinsey & Company, February 2018.

Burt, J. “IoT to Have Growing Impact on Businesses, Indus- tries, Survey Finds.” eWeek, May 4, 2016.

Chapter 13 • The Internet of Things as a Platform for Intelligent Applications 723

Chui, M., et al. “What It Takes to Get an Edge in the Internet of Things?” McKinsey Quarterly, September 2018.

Coumau, J., et al. “A Smart Home Is Where the Bot Is.” McKinsey Quarterly, January 2017.

Deichmann, J., M. Roggendorf, and D. Wee. “Preparing IT Sys- tems and Organizations for the Internet of Things.” McKin- sey & Company, November 2015.

Diaz, J. “CES 2017: LG’s New Smart Fridge Is Powered by Alexa.” Android Headlines, January 4, 2017. androidheadlines. com/2017/01/ces-2017-lgs-new-smar t-fr idge- powered-alexa.html/ (accessed August 2018).

Donaldson, J. “Is the Role of RFID in the Internet of Things Being Underestimated?” Mojix, May 2, 2017.

Durrios, J. “Four Ways IoT Is Driving Marketing Attribution.” Enterprise Innovation, April 8, 2017.

Editors. “Smart Cities Will Use 1.6B Connected Things in 2016.” eGov Innovation, December 22, 2015.

Editors. “Global Smart Cities IoT Technology Revenues to Exceed US$60 Billion by 2026.” Enterprise Innovation, January 23, 2018.

Estopace, E. “French National Railway Operator Taps IoT for Rail Safety.” eGov Innovation, February 21, 2017a.

Estopace, E. “Consortium to Build a Smart Mobility System for Hong Kong.” Enterprise Innovation, March 26, 2017b.

Fenwick, N. “IoT Devices Are Exploding on the Market.” Information Management, January 19, 2016.

Fitzgerald, M. “Data-Driven City Management: A Close Look at Amsterdam’s Smart City Initiative.” MIT Sloan Management Review, May 19, 2016.

Freeman, M. “Connected Cars: The Long Road to Autono- mous Vehicles.” San Diego Union Tribune, April 3, 2017.

Gemelli, M. “Smart Sensors Fulfilling the Promise of the IoT.” Sensors Magazine, October 13, 2017.

Greengard, S. “How AI Will Impact the Global Economy.” CIO Insight, October 7, 2016.

Hamblen, M. “Smart City Tech Connects Cars and Bikes with Big Data at MCW: Innovators Can Put Air Quality Sen- sors on Bicycles, While Wireless Connections Help Pave the Way for Driverless Cars.” Computerworld, February 22, 2016.

Hawkins, A. “Intel Is Working with Waymo to Build Fully Self- Driving Cars.” The Verge, September 18, 2017.

Hedge, Z. “Case Study: Athens International Airport Uses EXM and Libelium’s IoT Platform to Enhance Environmental Monitoring.” IoT Now, September 1, 2017.

Henderson, P. “10 Ways Analytics Can Make Your City Smarter.” InfoWorld and SAS Report AST = 0182248, June 6, 2017.

Hu, F. Security and Privacy in Internet of Things (IoTs): Mod- els, Algorithms, and Implementations. Boca Raton, FL: CRC Press, 2016.

Hupfer, S. “AI Is the Future of IoT.” IBM Blog, December 15, 2016. ibm.com/blogs/internet-of-things/ai-future-iot/ (accessed July 2018).

IBM. “Embracing the Internet of Things in the New Era of Cognitive Buildings.” White Paper. IBM Global Business Services, 2016.

Jamthe, S. The Internet of Things Business Primer. Stanford, CA: Sudha Jamthe, 2015.

Jamthe, S. IoT Disruptions 2020: Getting to the Connected World of 2020 with Deep Learning IoT. Seattle, WA: Create Space Independent Publishing Platform, 2016.

Kastrenakes, J. “Nest Can Now Use Your Phone to Tell When You’ve Left the House.” The Verge, March 10, 2016. theverge.com/2016/3/10/11188888/nest-now- uses-location-for-home-away-states-launches-family- accounts (accessed April 2018).

Khoury, A. “You Can Now Hail a Ride in a Fully Autonomous Ve- hicle, Courtesy of Waymo.” Digital Trends, February 17, 2018.

Korosec, K. “Toyota Is Using Nvidia’s Supercomputer to Bring Autonomous Driving to the Masses.” The Verge, May 10, 2017.

Koufopoulos, J. “9 Examples of the Internet of Things That Aren’t Nest.” Percolate, January 23, 2015.

Kvitka, C. “Navigate the Internet of Things.” January/February 2014. oracle.com/technetwork/issue-archive/2014/14- jan/o14interview-utzschneider-2074127.html (accessed April 2018).

Lacey, K. “Higher Ed Prepares for the Internet of Things.” Univer- sity Business, July 27, 2016. universitybusiness.com/arti- cle/higher-prepares-internet-things (accessed April 2018).

Libelium. “Smart Factory: Reducing Maintenance Costs and Ensuring Quality in the Manufacturing Process.” Libelium World, March 2, 2015. technology.ihs.com/531114/the- internet-of-everything-needs-a-fabric (accessed April 2018).

Manyika, J., M. Chui, P. Bisson, J. Woetzel, R. Dobbs, J. Bugh- in, and D. Aharon. “Unlocking the Potential of the Internet of Things.” McKinsey Global Institute, June 2015.

Marcus, J. “CNH Industrial Halves Product Downtime with IoT.” Product Lifecycle Report, May 6, 2015.

Martin, E. “AI May Have Your Health and Finances on Record Before the Year Is Out.” FutureFive, July 20, 2017. futurefive. co.nz/story/five-ways-ai-machine-will-affect-your-life- and-business-year/ (accessed April 2018).

McCafferty, D. “How the Internet of Things Is Changing Ev- erything.” Baseline, June 16, 2015.

McGrath, J. “Haier Wants You to Live Smaller and Smarter with Its New Appliances.” Digital Trends, January 5, 2016. digitaltrends.com/home/haier-shows-off-u-smart- appliances-at-ces-2016 (accessed April 2018).

McLellan, C. “Internet of Things in the Enterprise: The State of Play.” ZDNet.com, February 1, 2017a. zdnet.com/article/ enterprise-iot-in-2017-the-state-of-play/ (accessed April 2018).

McLellan, C. “Cybersecurity in an IoT and Mobile World.” Spe- cial Report. ZDNet, June 1, 2017b.

Meola, A. “What Is the Internet of Things (IoT)? Meaning & Definition.” Business Insider, May 10, 2018.

Miller, M. The Internet of Things: How Smart TVs, Smart Cars, Smart Homes, and Smart Cities Are Changing the World. Indianapolis, IN: Que Publishing, 2015.

Miller, R. “IoT Devices Could Be Next Customer Data Fron- tier.” TechCrunch, March 30, 2018.

724 Part IV • Robotics, Social Networks, AI and IoT

Morris, C. “Ordinary Home Appliances Are About to Get Really Sexy.” Fortune.com, January 6, 2016. fortune. com/2016/01/06/home-appliances-ces-2016 (accessed April 2018).

Morris, S., D. Griffin, and P. Gower. “Barclays Puts in Sensors to See Which Bankers Are Their Desks.” Bloomberg, Au- gust 18, 2017.

Murray, M. “Intel Lays Out Its Vision for a Fully Connected World.” PC Magazine, August 16, 2016.

Ohnsman, A. “Our Driverless Future Begins as Waymo Tran- sitions to Robot-Only Chauffeurs.” Forbes, November 7, 2017.

Park, H. “The Connected Customer: The Why Behind the In- ternet of Things.” Blue Hill Research. White Paper. January 2017.

Perkins, E. “Securing the Internet of Things.” Report G00300281. Gartner Inc., May 12, 2016.

Pitsker, K. “Put Smart Home Technologies to Work for You.” Kiplinger’s Personal Finance, October 2017.

PTC, Inc. “Internal Transformation for IoT Business Model Reshapes Connected Industrial Vehicles.” PTC Transfor- mational Case Study, November 12, 2015. ptc.com/~/ media/Files/PDFs/IoT/J6081_CNH_Industrial_Case _Study_Final_11-12-15.pdf?la=e (accessed April 2018).

Pujari, A. “Becoming a Smarter Manufacturer: How IoT Revo- lutionizes the Factory.” Enterprise Innovation, June 5, 2017.

Rainie, L., and J. Anderson. “The Internet of Things Connec- tivity Binge: What Are the Implications?” PewInternet.com, June 6, 2017.

SAS. “SAS Analytics for IoT: Smart Cities.” SAS White Paper 108482_G14942, September 2016.

SAS. “5 Steps for Turning Industrial IoT Data into a Competi- tive Advantage.” SAS White Paper 108670_G456z 0117. pdf, January 2017.

Scannell, B. “High Performance Inertial Sensors Propelling the Internet of Moving Things.” Technical Article. Analog De- vices, 2017.

Schwartz, S. Street Smart: The Rise of Cities and the Fall of Cars. Kindle Edition. New York, NY: Public Affairs, 2015.

Shah, S. “HPE, Tata to Build ‘World’s Largest’ IoT Network in India.” Internet of Business, February 27, 2017. internetof- business.com/hpe-tata-largest-iot-network-india/ (ac- cessed April 2018).

Sharda, R., et al. Business Intelligence, Analytics, and Data Science: A Managerial Perspective. 4th ed. New York, NY: Pearson, 2018.

Sinclair, B. IoT Inc.: How Your Company Can Use the Internet of Things to Win in the Outcome Economy. Kindle Edition, New York, NY: McGraw-Hill Education, 2017.

Solomon, S. “Israel Smart-Roads Startup Nabs Presti- gious  EY Journey Prize.” The Times of Israel, October 26, 2017.

Sorkin, A. “Larry Page’s Flying Taxis Now Exiting Stealth Mode.” The New York Times, March 12, 2018.

Staff. “Study Reveals Dramatic Increase in Capabilities for IoT Services.” Information Management, May 5, 2017.

Technavio. “Smart Sensors for the Fourth Industrial Revolu- tion: Molding the Future of Smart Industry with Advanced Technology.” Technavio.com, September 12, 2017.

Tokuoka, D. Emerging Technologies: Autonomous Cars. Ra- leigh, NC: Lulu.com, 2016.

Tomás, J. “Smart Factory Tech Defining the Future of Produc- tion Processes.” RCR Wireless News, March 28, 2016.

Townsend, A. Smart Cities: Big Data, Civic, Hackers and the Quest for a New Utopia. New York, NY: W. W. Norton, 2013.

Twentyman, J. “Athens International Airport Turns to IoT for Environmental Monitoring.” Internet of Business, Septem- ber 4, 2017.

Venkatakrishnan, K. “Are Connected Consumers Driving Smart Homes?” Enterprise Innovation, May 31, 2017.

Violino, B. “19 Top Paying Internet-of-Things Jobs.” Informa- tion Management, October 25, 2017.

Weinreich, A. “The Future of the Smart Home: Amazon, Walmart, & the Home That Shops for Itself.” Forbes, Febru- ary 1, 2018.

Weldon, D. “Steps for Getting an IoT Implementation Right.” Information Management, October 30, 2015.

725

Caveats of Analytics and AI

V P A R T

726

C H A P T E R

14 Implementation Issues: From Ethics and Privacy to Organizational and Societal Impacts

LEARNING OBJECTIVES

■■ Describe the major implementation issues of intelligent technologies

■■ Discuss legal, privacy and ethical issues ■■ Understand the deployment issues of intelligent systems

■■ Describe the major impacts on organizations and society

■■ Discuss and debate the impacts on jobs and work

■■ Discuss the arguments of utopia and dystopia in a debate of the future of robots and artificial intelligence (AI)

■■ Discuss the potential danger of mathematical models in analytics

■■ Describe the major influencing technology trends ■■ Describe the highlights of the future of intelligent systems

I n this concluding chapter, we cover a variety of issues related to the implementation and future of intelligent systems. We begin our coverage with technological issues such as security and connectivity. Then, we move to managerial issues that cover legality, privacy, and ethics. We next explore the impacts on organizations, society, work and jobs. Then, we review technology trends that point to the future.

Note. In this chapter we refer to all technologies covered in this book as intelligent technologies or intelligent systems.

This chapter has the following sections:

14.1 Opening Vignette: Why Did Uber Pay $245 Million to Waymo? 727 14.2 Implementing Intelligent Systems: An Overview 729 14.3 Legal, Privacy, and Ethical Issues 731 14.4 Successful Deployment of Intelligent Systems 737 14.5 Impacts of Intelligent Systems on Organizations 740 14.6 Impacts on Jobs and Work 747

Chapter 14 • Implementation Issues: From Ethics and Privacy to Organizational and Societal Impacts 727

14.7 Potential Dangers of Robots, AI, and Analytical Models 753 14.8 Relevant Technology Trends 756 14.9 Future of Intelligent Systems 760

14.1 OPENING VIGNETTE: Why Did Uber Pay $245 Million to Waymo?

In early 2018, Uber Technologies, Inc. paid $245 million worth of its own shares to Waymo Self-Driving Cars (a subsidiary of Alphabet). The payment was made to set- tle a lawsuit filed by Waymo alleging that Uber was using Waymo’s stolen proprietary technology.

THE BACKGROUND OF THE CASE

The lawsuit relates to the protection of intellectual property (trade secrets) owned by Waymo. As you may recall from Section 13.9, Waymo pioneered the self-driving car. A former engineer of Waymo (named Levandowski) allegedly illegally downloaded 14,000 of Waymo self-driving related confidential files. Worse than that, Levandowski may have convinced several top engineers of Waymo to leave Waymo and join him to create a start- up, Otto Company, for developing self-driving vehicles. Uber acquired Otto Company. For Uber, self-driving cars are essential for profitable growth when Uber will use such cars in a car-hailing system. Uber is a major car-hailing company that plans to move from sharing cars owned by individuals to the car-hailing business where self-driven cars will be owned by Uber and/or by car manufacturers. This way the profit for Uber could be much higher. Furthermore, Uber plans to operate driverless taxi fleets.

THE LEGAL DISPUTE

The legal dispute is very complicated. It deals with intellectual property and the ability of high technology employees to work after leaving their jobs for competitors.

Lawyers from Waymo claimed potentially huge damage if the Waymo trade secrets are used by the competitors. Waymo’s legal team based their case on a digital-forensics investigation that proved that Levandowski deliberately copied the confidential files and then tried to cover this downloading. Note that Uber did not steal trade secrets, but hired Levandowski, who had these secrets.

From a legal point of view, the case was unique, being the first related to self-driving cars, so there were no previous cases to rely on. The two companies are large tech com- panies in Silicon Valley.

Employees that leave companies are interviewed and reminded that they signed an agreement regarding trade secrets they acquired when working for the company they leave. Levandowski said in his exit interview from Waymo that his future plans did not include competing activities that may compete with Waymo’s self-driving cars. However, he had already met with Uber and sold it his new company, Otto Trucking. It became very clear that both Uber and Levandowski were not telling the truth.

WHY DID THEY SETTLE?

The rivals settled after four days in court. The case was in front of a jury, a fact that intro- duced an uncertain element to the case.

Waymo agreed to settle since, to win the case, it had to prove actual damage, which it was unable to do. Future damage is very difficult to compute. Furthermore, there was

728 Part V • Caveats of Analytics and AI

no evidence that Uber was using any of Waymo’s trade secrets, and Uber had already fired Levandowski.

Uber agreed to pay Waymo because the legal case constituted a possible delay in its development of self-driving cars, which is critical for the future of Uber. Also, legal fees were mounting (Uber is involved in several other legal issues related mostly to its driv- ers). Fighting Waymo did not ensure success given the deep pockets of Google. Actually, Waymo sent a clear message that it would protect its leading self-driving cars’ position at any cost.

CONCLUSION

• Uber paid about one-third of 1 percent in shares of its company. Uber was valued at $70 billion ( January 2018), which makes the payment equivalent to $245 million. Uber is planning on going public, which may increase its valuation.

• Uber agreed not to incorporate Waymo’s confidential information into its existing or future technology. This was a major condition of Waymo.

• The reason why this dispute is important to both companies is that the autonomous vehicle market could be worth $7 trillion by 2050 (per Marshall and Davies, 2018). Note: It is equal to about one-third of the total current U.S. national debt.

• There has been a major emerging change in the nature of the self-driving cars’ com- petition between 2016, when the legal dispute settled in July 2018.

Today, there are many more competitors and much more publicly known technologies and processes (i.e., fewer trade secrets). Finally, companies need to tell their employees what is not a trade secret.

u DISCUSSION QUESTIONS FOR THE OPENING VIGNETTE

1. Identify the legal issues involved in this case. 2. Why do you think Waymo agreed to take Uber’s shares instead of money? 3. What is the meaning of intellectual property in this case? 4. The presiding federal judge said at the end: “This case is now ancient history.” What

did he mean to say?

5. Summarize the potential damages to the two parties if they had continued with the legal dispute.

6. Summarize the benefits of the settlement to both sides.

WHAT CAN WE LEARN FROM THIS VIGNETTE

Self-driving cars are a major product of intelligent systems and artificial intelligence (AI) with huge potential benefits to its participants. Also, inevitable is the strong competition in the industry and the importance of trade secrets acquired along the way. Legal dis- putes are common in competitive situations, and the protection of intellectual property is critical. Intellectual property protection is one topic presented in our concluding chapter. Other issues that are related to the implementation of intelligent systems and are dis- cussed in this chapter are ethics, security, privacy, connectivity, integration, strategy, and top management roles.

We also learned in this vignette about the future importance of the new technology of autonomous vehicles. This technology may have a huge impact on organizations and their structure and operation. In addition, we discuss in this chapter the societal impacts of intelligent systems, and particularly their impact on work and jobs. We also explore some potential unintended consequences of intelligent systems. Finally, we explore the

Chapter 14 • Implementation Issues: From Ethics and Privacy to Organizational and Societal Impacts 729

potential future of intelligent systems and introduce the big debate regarding the dangers versus possible benefits of intelligent systems and particularly robots and AI.

Sources: Compiled from A. Marshall & A. Davies. (2018, February 9). “The End of Waymo v. Uber Marks a New Era for Self-Driving Cars: Reality.” Wired; A. Sage, et al. (2018, February 9). “Waymo Accepts $245 Million and Uber’s ‘Regret’ to Settle Self-Driving Car Dispute.” Reuters (Business News); K. Kokalitcheva. (2017, May 9). “The Full History of the Uber-Waymo Legal Fight.” Axios.

14.2 IMPLEMENTING INTELLIGENT SYSTEMS: AN OVERVIEW

Now that you have learned the essentials of analytics, data science, artificial intelligence, and decision support activities, you may be tempted to ask: What can I do with all this in my organization? You learned about the great benefits and you read about numerous companies that use intelligent systems. So, what you should do next? First read some of the resources recommended in this book so you will have a better understanding about the technologies. Next, read this chapter that deals with the major issues that are involved in implementing intelligent systems in organizations.

Implementing business analytic/AI systems can be a complex undertaking. In addition to specific issues found in intelligent systems, there are issues that are common to many other computer-based information systems. In this section, we describe the major types of issues, some of which are discussed in this chapter. For several success AI implementation factors revealed in a survey of 3000 executives, see Bughin, McCarthy, and Chui (2017).

The Intelligent Systems Implementation Process

This chapter is divided into three parts. In the first part, we describe some managerial- related implementation issues. In the second part, we describe the impacts of intelligent technologies on organizations, management, work, and jobs. The last part deals with technology trends and the future of intelligent technologies.

The implementation process of intelligent systems is similar to the generic process of other information systems. Therefore, we will present it only briefly. The process is illustrated in Figure 14.1.

THE MAJOR STEPS OF IMPLEMENTATION The major steps are:

Step 1 Need assessment. Need assessment needs to provide the business case for the intelligent systems, including their major parts. (This is a generic IT step and will not be discussed here.)

Step 2 Preparations. In this step, it is necessary to examine the organization readi- ness for analytics and AI. It is necessary to check available resources, employees’

Need Assessment

Preparation System

Development Deployment

Impact Assessment

Business Case Priority

Readiness status Available resources Employees’ attitude Legal privacy and ethics

Make or buy? Partnership

Security Integration

Success analysis Failure analysis Compare to targets Impact on people Impact on productivity

FIGURE 14.1 Implementation Process. Drawn by E. Turban

730 Part V • Caveats of Analytics and AI

attitudes for the change, projects’ priorities, and so on. This generic IT activity will not be discussed here. However, it is useful to think about legal, privacy, and ethi- cal issues as they are related to intelligent technologies as described in Section 14.3.

Step 3 System acquisition. Organizations need to decide on in-house or outsourc- ing approach (make or buy) or on a combination of the two and possibly with part- nership with a vendor or another company. A consultant may help at this step. It is a generic IT step that will not discussed here.

Step 4 System development. Regardless of who will develop the system, certain ac- tivities need to be done. These include security, integration with other systems, proj- ect management preparation, and other activities. Again, many of those are generic and will not be described here. Only selected ones are described in Section 14.4.

Step 5 Impact assessment. It is necessary to check the performance of the systems against plans. Again, this is a generic issue that will not be covered here.

The Impacts of Intelligent Systems

Intelligent systems are impacting all our lives and many businesses and other organiza- tions. It is much easier to find what is not impacted than what is impacted. In this section, we divide these impacts into three categories as shown in Figure 14.2 with the section numbers where they are presented. We exclude from this list the impact on individuals and quality of life, which is a very large field (health, education, entertainment, crime fighting, social services, etc.).

Example

Here is the example in the entertainment field. In the near future, when you go Disneyland, Disney World, or one of the Disney International Parks, you will see high- flying acrobatic robots. You will see them everywhere there, and it is amazing. For a pre- view, watch the following videos: money.cnn.com/video/news/2018/07/04/disney- robots-acrobatics-stuntronics-animatronics.cnnmoney/index.html and youtube. com/watch?v=Z_QGsNpI0J8.

Impacts of Intelligent Systems

Impacts on Organization Work and Jobs

Section 14.5 Section 14.6

Potential Unintended Impacts

Section 14.7

Structure Employees Management & D/M Industries Competition

Jobs to be automated Safe jobs Changes on the nature of work

Dangers of AI and robots Dangers of analytical model Mitigating the dangers

FIGURE 14.2 Impact Landscape. Drawn by E. Turban

Chapter 14 • Implementation Issues: From Ethics and Privacy to Organizational and Societal Impacts 731

u SECTION 14.2 REVIEW QUESTIONS

1. List the major steps in the implementation process. 2. Why is implementation an important subject? 3. Describe the major impact areas of intelligent systems.

14.3 LEGAL, PRIVACY, AND ETHICAL ISSUES

As data science, analytics, cognitive computing, and AI grow in reach and pervasive- ness, everyone may be affected by these applications. Just because something is doable through technology does not make it appropriate, legal, or ethical. Data science and AI professionals and managers have to be very aware of these concerns. Several important legal, privacy, and ethical issues are related to intelligent technologies and they are inter- related. For example, several privacy issues are parts of ethics or have legal aspects. Here we provide only representative examples and sources as pointed out in Chapter 1. Our goal here is only to give the reader an exposure to these issues. For why should we care about the legal, ethical, and privacy of AI, see Krigsman (2017).

Legal Issues

The introduction of intelligent technologies may compound a host of legal issues already relevant to computer systems. For example, questions concerning liability for the actions of advice provided by intelligent machines are beginning to be considered. In this sec- tion, we provide a sample of representative issues. Many more exist.

In addition to resolving disputes about the unexpected and possibly damaging re- sults of some intelligent systems (see the opening vignette and Section 14.7), other com- plex issues may surface. For example, who is liable if an enterprise finds itself bankrupt as a result of using the advice of an AI-based application? Will the enterprise itself be held responsible for not testing the system adequately before entrusting it with sensitive or volatile issues? Will auditing and accounting firms share the liability for failing to apply adequate auditing tests? Will the software developers of intelligent systems be jointly liable? As self-driving cars become more common, who is liable for any damage or injury when a car’s sensors, network, or AI system fail to function as planned? A recent case involving a Tesla car accident where the driver died in a crash while the car was alleg- edly on “autopilot” mode has brought this issue to the front pages of newspapers and the legal profession.

A SAMPLE OF AI POTENTIAL LEGAL ISSUES • What is the value of an expert opinion in court when the expertise is encoded in a

computer? • Who is liable for wrong advice (or information) provided by an intelligent application?

For example, what happens if a physician accepts an incorrect diagnosis made by a computer and performs a procedure that results in the death of a patient?

• What happens if a manager enters an incorrect judgment value into an intelligent ap- plication and the result is damage or a disaster?

• Who owns the knowledge in a knowledge base (e.g., the knowledge of a chatbot)? • Can management force experts to contribute their expertise to an intelligent system?

How will they be compensated? • Is it okay for self-driving cars with in-vehicle back-up drivers to drive on public

roads? (Yes, in a few states, notably in California.) • Who should regulate driverless car: cities, states, or the federal government? • U.S. federal regulators are creating national laws for self-driving cars (for safe driving).

732 Part V • Caveats of Analytics and AI

• Should delivery robots be allowed on sidewalks? (Not in San Francisco but in some European cities)

• Are drivers of Uber and similar companies self-employed? (Not in London, the United Kingdom)

• Should robots have human rights? (What if they are citizens like Sophia in Saudi Arabia?) If they get rights, should they have legal responsibilities as well?

• Should we legalize robot taxis? Would this make trips cheaper? (Yes in Singapore and other places, and it can be cheaper)

Source: Turban, Introduction to Information Technology, 2nd edition, John Wiley & Sons, 2006.

Example: Intellectual Property Protection

The opening vignette directed our attention to a legal issue that is very important for technology-related companies: the ownership and protection of intellectual property.

LEGAL ISSUES OF INTELLIGENT TECHNOLOGIES Several of the ethical issues described later need to be combined with legal issues. For example, take robots’ legal rights. Do we need these rights? What for (an ethical issue)? Then, it will be necessary to develop the legal rights. Facebook, for example, has had legal issues regarding face recognition. Safety rules for robots were developed a long time ago. At the moment, there are very few laws regarding intelligent technologies. Most of the laws relate to safety.

AI AND LAW In addition to laws related to robotics and AI, there is a subfield of AI that is concerned with AI applications to the legal profession and the solution of some legal problems. According to Donahue (2018), the following are some major topics:

• Analyzing legal-related data (e.g., regulatory conflicts) to detect pattern • Providing legal advice to consumers (e.g., see DoNotPay.com). • Document review • Analyzing contracts • Supporting legal research • Predicting results (e.g., likelihood to win) • AI impact on the legal profession.

AI can execute routine legal-related tasks such as managing documents and drafting contracts. For details, see Kahn (2017). For 35 applications in law and legal practice see Rayo (2018). Legal issues may be strongly connected to our next topic, privacy.

Privacy Issues

Privacy means different things to different people. In general, privacy is the right to be left alone and the right to be free from unreasonable personal intrusions. Privacy has long been related to legal, ethical, and social issues in many countries. The right to privacy is recognized today in every state of the United States and by the federal government either by statute or by common law. The definition of privacy can be interpreted quite broadly. However, the following two rules have been followed fairly closely in past court decisions: (1) The right of privacy is not absolute. Privacy must be balanced against the needs of society. (2) The public’s right to know is superior to the individual’s right to privacy. These two rules show why it is difficult, in some cases, to determine and enforce privacy regulations. Privacy issues online have specific characteristics and policies. One area where privacy may be jeopardized is discussed next. Privacy issues are getting more

Chapter 14 • Implementation Issues: From Ethics and Privacy to Organizational and Societal Impacts 733

and more important as the amount of data generated on the Internet is increasing expo- nentially, and in many cases it is lightly secured. For an overview of privacy as it relates to AI, see Provazza (2017).

COLLECTING INFORMATION ABOUT INDIVIDUALS Intelligent technologies aim to pro- vide targeted services and marketing to consumers; they do so by collecting informa- tion about these customers. In the past, the complexity of collecting, sorting, filing, and accessing information manually from numerous government agencies and other public databases was, in many cases, a built-in protection against the misuse of private informa- tion. The Internet in combination with large-scale databases has created an entirely new dimension of accessing and using data. The inherent power in intelligent systems that can access vast amounts of data and interpret them can be used for the good of society. For example, by analyzing records with the aid of business analysis, it is possible to elimi- nate or reduce fraud, crime, government mismanagement, tax evasion, welfare cheating, family-support filching, employment of illegal workers, and so on. However, what price must the individual pay in terms of loss of privacy so that the government can better apprehend criminals? The same is true on the corporate level. Private information about employees may aid in better corporate decision making, but the employees’ privacy may be compromised.

The use of AI technologies in the administration and enforcement of laws and regu- lations may increase public concern regarding privacy of information. These fears, gener- ated by the perceived abilities of AI, will have to be addressed at the outset of almost any AI development effort.

VIRTUAL PERSONAL ASSISTANTS Amazon’s Echo/Alexa and similar devices listen to what is going on. They also may take photos. In other words, your voice assistant is spy- ing on you.

Most advanced is the Echo/Alexa pair. You can ask Alexa to buy Amazon products today. Amazon and Google filed for a patent that will enable the virtual assistants in your home to advertise and sell you products. Privacy advocates are not happy, but customers may be. For example, Elgen (2017) describe how Alexa acts as a fashion consultant, using style check. The system combines the knowledge of a fashion specialist and AI knowl- edge. A recommendation provides you with two photos at a time, telling you which one to buy (based on color, current trends, etc.). To make it useful, Amazon is improving the privacy. This may not be easy since your record is stored in Amazon’s cloud.

Huff (2017) provides arguments about the risks of the assistant and the protection provided by Amazon.

MOBILE USER PRIVACY Many users are unaware of the private information being tracked through their smartphone usage. Many apps collect user data that track each phone call as it moves from one cell tower to another, from GPS-enabled devices that transmit users’ locations, and from phones transmitting information at Wi-Fi hotspots. Major app developers claim that they are extremely careful and protective of users’ privacy, but it is interesting to note how much information is available through the use of a single device, especially when smartphones contain more and more AI components.

PRIVACY IN IOT NETWORKS For privacy and security of the Internet of Things (IoT), see Hu (2016). More data are flowing with IoT networks. Note that AI data privacy issues are on the rise, especially when AI deals with consumers’ data. There is a growing amount of data collected, for example, by machine learning and chatbots. Also, in the enterprise, employers collect and analyze more data on employees. How do we protect the data and guard against their misuse?

734 Part V • Caveats of Analytics and AI

RECENT TECHNOLOGY ISSUES IN PRIVACY AND ANALYTICS With the growth of Internet users in general and mobile device users in particular, many companies have started to employ intelligent technologies to develop profiles of users on the basis of their device usage, surfing, and contacts. The Wall Street Journal has an excellent collection of articles titled “What They Know” (WallStreetJournal.com, 2016). These articles are constantly updated to highlight the latest technology and privacy/ethical issues. One of the companies mentioned in this series is Rapleaf (now part of Towerdata). Rapleaf’s technology claims to be able to provide a profile of a user just by knowing his or her e-mail address. Clearly, Rapleaf’s technology enables it to gather significant related information. Another company that aims to identify devices on the basis of their usage is BlueCava, which recently merged with Qualia (Qualia.com). Qualia’s BlueCava technology attaches a personal profile to be able to recognize a user as one individual or a household, even though the user may be working with multiple mo- bile devices and laptops. All of these companies employ analytics such as clustering and association mining to develop profiles of users. Of course, many of the analytics start-ups in this space claim to honor user privacy, but violations are often reported. For example, Rapleaf was collecting unauthorized information from Facebook users and was subsequently banned from Facebook. One user reported that an hour after he gave his e-mail address to a company that specializes in user information monitoring (reputation.com), the company was able to discover his Social Security number. So, violations of privacy create fears of criminal conduct regarding information. This area is a big concern overall and needs careful study. These examples not only illustrate the power of analytics in being able to learn more about target customers but also serve as a warning to AI and analytics professionals about being sensitive to privacy and ethical issues.

Another related application area of privacy concerns is analyzing employee be- haviors on the basis of data collected from sensors that employees wear in a badge. One company, Humanyze, has reported several such applications of its sensor-embedded badges. These sensors track all movements of an employee.

Example: Using Sensors and IoT to Observe Bankers at Barclays Bank

Using heat and motion sensors, Barclays tracks how long its bankers are at their desks. The system was installed in the London, United Kingdom, branches. The formal explanation was to find out the occupancy of the cubes in the bank to optimally allocate and possibly reduce office space. The IoT network provided dashboards showing which workstations (cubes) were underutilized, and what the usage trend was. The bank informed the em- ployees and the union that this project did not measure productivity, only space utiliza- tion. The results can be used to better manage energy consumption in the cubes and to schedule a flexible work environment. As a result, Barclays was able to save office space and rent it out for $45 million a year.

The bank uses a similar tracking system to find out how much time that different types of employees spend with customers. The union is watching this IoT application carefully to ensure that it is not used to spy on employees. Other banks in England use similar systems. For details, see Bloomberg News (2017).

Of course, situations like those described create major privacy concerns. Should companies be able to monitor their employees this intrusively?

Finally, there is a possibility of ransomware, or hackers’ attacks on robots, which could be used against businesses whose employees use such robots. Smith (2018) re- ported on research that identified 50 vulnerabilities in robots. Ransomware attacks may interrupt operations, forcing organizations to pay substantial ransoms.

Chapter 14 • Implementation Issues: From Ethics and Privacy to Organizational and Societal Impacts 735

OTHER ISSUES OF POTENTIAL PRIVACY VIOLATION The following are some more ex- amples of potential privacy violations in the intelligent technology world:

• Delaware police are using AI dashcoms to look for fugitives in passing cars. Photos and videos taken are sent to the clouds and analyzed there by AI algorithms.

• Facebook’s face recognition systems create concerns regarding privacy protection. • Epicenter offers its employees a microchip implant. It acts like a swipe card, opens

doors, buys you food in the company store, and much more. But management can track you too. It is given only to volunteers.

Who Owns Our Private Data?

With the recent growth of data from our use of technology and the companies’ abil- ity to access and mine it, the privacy debate also leads to the obvious issue of whose property any user’s data is; see Welch (2016) for highlights in this issue in a Bloomberg Businessweek column. Take an example of a relatively new car. The car is equipped with many sensors starting with tire pressure sensors to GPS trackers that can keep track of where you have gone, how fast you were driving, when you changed lanes, and so on. The car may even know the passenger’s weight added to the front seat. As Welch notes, a car connected to the Internet (most new cars are!) can be a privacy nightmare for the owner or a data “gold mine” for whoever can possess or analyze these data. A major battle is brewing between automobile manufacturers and technology providers such as Apple (CarPlay) and Google (Android Auto) on who owns these data and who can ac- cess them. This is becoming more crucial because as cars become smarter and eventu- ally self-driving, the driver/passenger in the car could be a highly targeted prospect for marketers’ services. For example, Google’s Waze app collects GPS data from millions of users to track traffic information and help users find the best routes; but it also displays pop-up ads on the users’ screens. Yelp, Spotify, and other apps popularly used in cars have similar approaches.

The bottom line is that intelligent systems professionals and users must be aware of the legal and ethical issues involved in collecting information that may be privileged or protected. Privacy issues are considered in many cases as important components of ethics.

Ethics Issues

Several ethical issues are related to intelligent systems. Personal values constitute a major factor in the issue of ethical decision making. The study of ethical issues is complex be- cause of their multidimensional nature. One story that upset many users (although it was not illegal) some time ago was Facebook’s experiment to present different News Feeds to the users and monitor their emotional reactions as measured by replies, likes, senti- ment analysis, and so on. Most companies, including technology companies, run user testing to identify the features most liked or disliked and fine-tune their product offerings accordingly. Because Facebook is so large, running this experiment without the users’ informed consent was viewed as unethical. Indeed, Facebook acknowledged its error and instituted a more formal review through Internal Review Boards and other compliance mechanisms for future testing.

Morgan (2017) said that it is necessary to be at the foundations of what AI does for both vendors and customers in order to stay ethical and have transparency of each situation. This way people can stay honest and adhere to the goals of AI, so it can play a significant role in our life and work. For how ethical issues interfere with Alphabet’s (Google) initiatives, see Kahn (2017).

736 Part V • Caveats of Analytics and AI

Ethical Issues of Intelligent Systems

Many people have raised questions regarding ethical issues in AI, robotics, and other in- telligent systems. For example, Bossmann (2016) raised the following issues:

1. What are their impact on jobs (see Section 14.5)? 2. How do machines (i.e., robots) affect our behavior and interactions? 3. How can wealth created by intelligent machines be distributed (e.g., Kaplan, 2016)? 4. How can intelligent applications mistakes be guarded against? For example, how

long should training programs in machine learning be? 5. Can intelligent systems be fair and unbiased? How can bias in creation and operation

of AI systems be eliminated? 6. How can intelligent applications be keep safe from adversaries? 7. How can systems be protected against unintended consequences (e.g., accidents in

robot operations)? For example, Facebook researchers had to shut down an AI sys- tem that created its own poor language.

8. How can we stay in control of a complex intelligent system? 9. Should we develop robots’ legal rights? How can we define and plan human treat-

ment of intelligent machines? 10. Should we allow a self-governing robot society to exist with ours? 11. To what extent should we influence unintended robots’ behavior (or even be able to)? 12. How would we get around the question of smart machine ownership?

Additional issues are:

• Electronic surveillance. • Ethics in business intelligence (BI) and AI systems design. • Software piracy. • Invasion of individuals’ privacy. • Use of proprietary databases and knowledge bases. • Use of personal intellectual property such as knowledge and expertise for the

benefits of companies and the payment to the contributors. • Accuracy of data, information, and knowledge. • Protection of the rights of users. • Accessibility to information by AI users. • The amount of decision making to delegate to intelligent machines.

• How AI can fail due to inappropriate ethics. • The ethics of legal analytics (Goldman, 2018).

Other Topics in Intelligent Systems Ethics

• Machine ethics is a part of the ethics of AI that is concerned with the moral behavior of artificially intelligent beings (per Wikipedia; see details there).

• Robotics is concerned with the moral behavior of designers, builders, and users of robots.

• Microsoft’s Tay chatbot was closed due to its inability to understand many irrelevant and offending comments.

• Some are afraid that algorithm-based technologies, including AI, may become rac- ists. We discuss this topic in Section 14.8. Also, see Clozel (2017).

• According to Spangler (2017), self-driving cars may one day face a decision of whom to save and whom to kill.

• Voice technologies enable the identification of callers to AI machines. This may be great on one hand, but it creates privacy concerns on the other.

Chapter 14 • Implementation Issues: From Ethics and Privacy to Organizational and Societal Impacts 737

• One area in which there are considerable ethical concerns (frequently combined with legal concerns) is the healthcare/medical field. Given the large efforts by Alphabet and IBM Watson initiatives, this is not surprising. For a discussion, see Bloomberg News (2017).

For comprehensive coverage of ethical issues in big data and data sharing, see Anon (2017). For principles for Big Data analysis, see Kassner (2017).

COMPUTER ETHICS IN GENERAL Computer ethics focuses on the behavior of people toward information systems and computers in general. The study of ethics in intelligent systems is strongly related to the ethics of computers and information systems in general. The following are some resources.

THE TEN COMMANDMENTS OF COMPUTER ETHICS This well-known document is pub- lished by cybercitizenship (cybercitizenship.org/ethics/commandments.html).

1. Thou shalt not use a computer to harm other people. 2. Thou shalt not interfere with other people’s computer work. 3. Thou shalt not snoop around in other people’s files. 4. Thou shalt not use a computer to steal. 5. Thou shalt not use a computer to bear false witness. 6. Thou shalt not use or copy software for which you have not paid. 7. Thou shalt not use other people’s computer resources without authorization. 8. Thou shalt not appropriate other people’s intellectual output. 9. Thou shalt not think about the social consequences of the program you write.

10. Thou shalt not use a computer in ways that show consideration and respect.

A major upcoming issue is that of ethics for autonomous vehicles. For example, who will develop them, how will they be programmed into the vehicles, and how will they be enforced? See Sharma (2017).

For review of ethical issue considerations in information research literature, see nowpublishers.com/article/Details/ISY-012/.

MIT Media Lab and the Center for Internet & Society at Harvard University manage an initiative to research ethical and governance topics in AI. SAS, a major analytical and AI vendor, proposed three essential steps for AI ethics as described in sas.com/en_us/ insights/articles/analytics/artificial-intelligence-ethics.html/.

u SECTION 14.3 REVIEW QUESTIONS

1. List some legal issues of intelligent systems. 2. Describe privacy concerns in intelligent systems. 3. In your view, who should own the data about your use of a car? Why? 4. List ethical issues in intelligent systems. 5. What are the 10 commandments of computer/information systems?

14.4 SUCCESSFUL DEPLOYMENT OF INTELLIGENT SYSTEMS

Many experts, consultants, and researchers provide suggestions regarding intelligent sys- tems’ successful deployment. Given the importance of the topic, it is clear that companies need to get ready for the mass arrival of AI and other intelligent technologies. Here are some topics related to deployment strategy:

• When to embark on intelligent projects and how to prioritize them. • How to decide whether to do it yourself or use partners, or to outsource.

738 Part V • Caveats of Analytics and AI

For more strategy issues, see Kiron (2017). We cover only several topics in this section and provide references to more. Most of the implementation topics are generic in nature and will not be covered here.

Top Management and Implementation

According to Chui et al. (2017), from McKinsey & Company, “Senior executives need to understand the tactical as well as the strategic opportunities (of AI), redesign their organi- zations, and commit to helping shape and debate about the future of work.” Specifically, the executives need to plan for integrating intelligent systems into their workplace, making a commitment to conduct a participating environment for the changes and provide suf- ficient resources. Snyder (2017) claims that many executives know that intelligent systems will change their business, but they do not do much about it.

KPMG, a large management service consultant, provides the following steps regard- ing digital labor:

“KPMG’s holistic approach-from strategy through execution will assist compa- nies on each step of implementation. The steps are:

• Establishing priority areas for technological innovation. • Developing a strategy and a plan for the employees. • Identify providers and partners for plans’ execution. • Establishing a strategy and plans to realize benefits from the digital labor initiatives.”

Source: KPMG Internal Audit: Top 10 in 2018, Considerations for impactful internal audit departments, © 2018 KPMG LLP.

A complete guide for KPMG is provided by Kiron (2017). It includes robotic process au- tomation, enhanced process automation, and cognitive automation. For issues regarding leadership in implementation, see Ainsworth (2017).

System Development Implementation Issues

Since AI and business analytics are broad terms, describing several technologies whose maturity levels vary, implementation issues may vary considerably. Shchutskaya (2017) cites the following three major problems:

1. Development approach. Business analytic and AI systems require an approach dif- ferent from that of other IT/computer systems. Specifically, it is necessary to identify and deal with different and frequently large data sources (see the opening vignette to Chapters 1 and 2). It is necessary to cleanse and curate these data. Also, if learning is involved, one needs to use machine training. Thus, special methodologies are needed.

2. Learning from data. Many AI and business analytics involve learning. The quality of the input data determines the quality of the applications. Also, the learning mech- anism is important. Therefore, data accuracy is critical. In learning, systems must be

• How to justify investments in intelligent projects. • How to overcome employees’ resistance (e.g., fear of job loss). • How to arrange appropriate people-robot teams. • How to determine which decisions to fully automate by AI. • How to protect intelligent systems (security) and how to protect privacy. • How to handle possible loss of jobs and retraining of employees (Section 14.5). • How to determine whether you have the necessary up-to-date technology. • How to decide what support top management should provide. • How to integrate the system with business processes. • How to find qualified personnel for building and using intelligent systems.

Chapter 14 • Implementation Issues: From Ethics and Privacy to Organizational and Societal Impacts 739

able to deal with changing environmental conditions. Data should be organized in databases, not in files.

3. No clear view is available of how insights are generated. AI, IoT, and business ana- lytic systems generate insights, conclusions, and recommendations based on the analysis of the data collected. Given that data are frequently collected by sensors and there are different types of them, we may not have a clear view of the insights that are generated.

Related important areas include problems with Big Data, ineffective information access, and limited integration capabilities (discussed next).

Connectivity and Integration

As part of the development process, it is necessary to connect the AI and analytic applica- tions to existing IT systems, including the Internet, and other intelligent systems.

Example

The Australian government commissioned Microsoft in August 2017 to build hyperscale cloud regions to unlock the power of intelligent technologies. The system is expected to dramatically modernize how the government processes data and delivers services to its citizens. The system can handle both unclassified and protected data. The infrastructure is built inside, or near, the government data centers. The system will enable the gov- ernment to use innovative applications based on machine learning, bots, and language translation, and it will improve healthcare, education, social services, and other govern- ment operations. Finally, the system will increase both security and privacy protection.

Integration needs to be done with almost every system that is being impacted by AI or business analytic. For example, it is necessary to integrate intelligent applications both to a digital marketing strategy and to marketing implementation. For a discussion, see searchenginejournal.com/artificial-intelligence-marketing/200852/.

To overcome the integration difficulty, Huawei of China (a cellphone producer) is installing an AI system with its knowledge base inside the chips of its products. Other phones’ manufacturers rely on connecting to the “cloud” to interact there with AI knowledge. For the implications on IoT connectivity, see Rainie and Anderson (2017). For considerations regarding IoT connectivity providers, see Baroudy et al. (2018).

Security Protection

Many intelligent applications are managed and updated in the “cloud” and/or connected to the regular Internet. Unfortunately, by adding Internet connection, new vulnerabilities may be created. Hackers use intelligent technologies to identify these vulnerabilities. For how criminals use AI and related issues, see Crosman (2017). In Section 14.7, we discuss the potential dangers of robotics. The safety of passengers in self-driving cars and others who may be involved in collisions with the self-driving cars is an important safety issue as well. Also, the safety of people working near robots has been researched for many decades. In addition, hacking robots, chatbots, and other intelligent systems are areas that require attention. Finally, the safety of robots themselves when they work on the streets is an issue. Some people attack them (see McFarland, 2017a and the video there).

Leveraging Intelligent Systems in Business

There are many ways to leverage intelligent systems, depending on the nature of the applications. Catliff (2017) suggests the following ways to do this, leveraging the

740 Part V • Caveats of Analytics and AI

intelligent technology capabilities to increase efficiency and provide more customer care. Specifically, he suggested:

1. Customize the customer experience (e.g., for interactions with customers). 2. Increase customer engagement (e.g., via chatbots). 3. Use intelligent technologies to detect problems and anomalies in data.

Singh (2017a) recommends the following as critical success factors: discover, predict, justify, and learn from experience. Ross (2017) raised the issue of the need to upgrade employees’ skills and build an empowered AI-savvy workforce. One of the most important issues is how to handle the fear of job loss of employees. This is discussed in Section 14.6.

Intelligent System Adoption

Most of the issues related to intelligent systems’ adoption are the same as or similar to that of any information systems. For example, employees may resist change, management may not provide sufficient resources, there could be a lack of planning and coordina- tion, and so on. To deal with such issues, Morgan Stanley drew ideas from hundreds of conversations with experts (see DiCamillo, 2018). One important issue is to have an ap- propriate deployment and adoption strategy that should work in harmony with the imple- mented technologies and the people involved. In general, the generic adoption approach to information systems should work here, too.

u SECTION 14.4 REVIEW QUESTIONS

1. Describe the systems deployment process. 2. Discuss the role of top management in deploying intelligent systems. 3. Why is connectivity such an important issue? 4. Describe system development issues. 5. Discuss the importance of security and safety, and how to protect them. 6. Describe some issues in intelligent systems adoption.

14.5 IMPACTS OF INTELLIGENT SYSTEMS ON ORGANIZATIONS

Intelligent systems are important components in the information and knowledge revolu- tion. Unlike the slower revolutions of the past, such as the Industrial Revolution, this revolution is taking place very rapidly and affecting every facet of our work and lives. Inherent in this transformation is the impact on organizations, industries, and managers, some of which are described in this section.

Separating the impact of intelligent systems from that of other computerized systems is a difficult task, especially because of the trend toward integrating, or even embedding, intelligent systems with other computer-based information systems. Intelligent systems can have both micro- and macro implications. Such systems can affect particular individu- als and jobs as well as the work and structures of departments and units within an organi- zation. They can also have significant long-term effects on total organizational structures, entire industries, communities, and society as a whole (i.e., regarding macro impact, see Sections 14.6 and 14.7).

Explosive growth in analytics, AI, and cognitive computing is going to have a major impact on the future of organizations. The impact of computers and intelligent systems can be divided into three general categories: organizational, individual, and societal. In each of these, computers may have many possible impacts. We cannot possibly consider all of them in this book, so in the next paragraphs we cover topics we feel are most rel- evant to intelligent systems and organizations.

Chapter 14 • Implementation Issues: From Ethics and Privacy to Organizational and Societal Impacts 741

New Organizational Units and Their Management

One change in organizational structure is the possibility of creating an analytics depart- ment, a BI department, a data science department, and/or an AI department in which analytics plays a major role. Such special units (of any type) can be combined with or replace a quantitative analysis unit, or it can be a completely new entity. Some large cor- porations have separate decision support units or departments. For example, many major banks have such departments in their financial services divisions. Many companies have small data science or BI/data warehouse units. These types of departments are usually in- volved in training in addition to consulting and application development activities. Others have empowered a chief technology officer over BI, intelligent systems, and e-commerce applications. Companies such as Target and Walmart have major investments in such units, which are constantly analyzing their data to determine the efficiency of marketing and supply chain management by understanding their customer and supplier interactions. On the other hand, many companies are embedding analytics/data science specialties within functional areas such as marketing, finance, and operations. In general, this is one area where considerable job opportunities currently exist. For a discussion of the need for a chief data officer, see Weldon (2018). Also, the need for a chief AI officer is discussed by Lawson (2017).

Growth of the BI and analytics has resulted in the formation of new units within IT companies as well. For example, a few years ago, IBM formed a new business unit focused on analytics. This group includes units in BI, optimization models, data mining, and business performance. More importantly, the group is focused not just on software but also significantly more on services/consulting.

Transforming Businesses and Increasing Competitive Advantage

One of the major impacts of intelligent systems is the transformation of businesses to digi- tal ones. While such transformation has been going on with other information technolo- gies for years, it has accelerated with intelligent technologies, mostly with AI.

In many cases, AI is only a supportive tool for humans. However, as AI has become more capable, machines have been able to perform more tasks by themselves or with people. The fact is that AI already is transforming some businesses. As seen in Chapter 2, AI already is changing all business functional areas, especially marketing and finance. The impact ranges from full automation of many tasks, including managerial ones, to an increase in human-machine collaboration (Chapter 11). A comprehensive description of how AI is driving digital transformation is provided by Daugherty and Wilson (2018), who concluded that businesses that will miss the AI-driven transformation would be in a competitive disadvantage. Batra et al. (2018) point to a similar phenomenon and urge companies to use AI and utilize it for a wave of innovations. For more on this topic, see Uzialko (2017).

USING INTELLIGENT SYSTEMS TO GAIN COMPETITIVE ADVANTAGE Use of intelligent technologies, and especially AI, is evidenced in many cases. For example, using robots, Amazon.com enabled the company to reduce cost and control online commerce. In gen- eral, by cutting costs, increasing customer experiences, improving quality, and speeding deliveries, companies will gain competitive advantage. Rikert (2017) describes conversa- tions with CEOs about how AI and machine learning can beat the competitors. Andronic (2017) points to competitive advantage. The benefits include generating more demand (see Chapter 2), automating sales (Chapter 2), and identifying sales opportunities.

An important recent factor is the fact that new companies and blurring sector bor- ders are influencing the competitive picture of many industries. For example, autono- mous vehicles will impact the competition in the automotive industry.

742 Part V • Caveats of Analytics and AI

According to Weldon (2017c), a smart use of analytics offers top competitive ad- vantage. The author provides advice on how organizations can get the full benefits from analytics. An example of how 1-800-Flowers.com is using analytics, AI, and other intel- ligent technologies to gain a competitive advantage is provided in Application Case 14.1.

1-800-Flowers.com is a leading online retailer of flowers and gifts. The company moved from tele- phone to online ordering in the mid-1990s. Since then, it has grown to over $1 billion in revenue and over 4,000 employees, despite fierce competi- tion. In a world dominated by online giants such as Amazon.com and Walmart.com, and hundreds of other companies that sell online flowers and gifts, survival is not easy.

The company is using the following three key strategies:

• Enhancing the customer experience. • Driving demand more efficiently. • Building a workforce that supports the

products and technology innovation (culture of innovation).

The company has been using intelligent technolo- gies extensively to build a superb supply chain and to facilitate collaboration. Lately, it started to use intelligent systems to enhance its competitive strate- gies. Here are several technologies covered in this book that the company uses.

1. Optimal customers experience. Using SAS Marketing Automation and Data Manage- ment products, the retailer collects informa- tion regarding customers’ needs and analyzes it. This information enables senders of flow- ers and gifts to find perfect gifts for any occa- sion. Senders want to make recipients happy, so appropriate recommendations are critical. The company uses advanced analytics and data mining from SAS to anticipate custom- ers’ needs. 1-800-Flowers.com marketers can then communicate with customers more effec- tively. Using the newest tools, company data scientists and marketing analysts mine data more efficiently. Today customer expectations are higher than ever because it is much easier for customers to compare vendors’ offerings

online. Analytics and AI enable the company to understand its customers’ sentiments. Now the company is able to understand the emo- tional reasoning behavior for purchasing deci- sions and customer loyalty. This change results in product recommendations described later.

2. Chatbots. 1-800-Flowers.com has a bot on Facebook Messenger. As described in Chapter 12, such a bot can be useful as a source of informa- tion and as a vehicle for conversation. The com- pany also offers chat on its Web site online, and chat using voice. In addition, mobile shoppers can use Google Assistant for voice ordering. The company also offers voice-enabled Alexa with its “one-shot intent” to expedite ordering.

3. Customer service. The company offers a portal and one-stop shopping similar to what Amazon.com offers, and self-service payment is available. The same capability is available when shopping with the company’s bot on Facebook Messenger. Customers do not have to leave Facebook to complete an order.

4. AI-based recommendation. As you may recall from Chapter 12, e-commerce retailers excel by providing product recommendation (e.g., Amazon, Netflix). 1-800-Flowers.com is do- ing the same thing, offering recommendation and advice on gifts from their brand’s websites (e.g., Harry and David). The recommendations are generated by IBM’s Watson and are offered as a “cognitive concierge,” making online shop- ping feel as having an in-store experience. This AI-based service is known as GWYN (Gifts When You Need) at 1-800-flowers. Watson’s natural language processing (NLP) enables easy shopper-machine conversations.

5. Personalization. SAS advanced analytics en- ables the company’s marketing department to segment customers into groups with similar characteristics. Then the company can send

Application Case 14.1 How 1-800-Flowers.com Uses Intelligent Systems for Competitive Advantage

Chapter 14 • Implementation Issues: From Ethics and Privacy to Organizational and Societal Impacts 743

Redesign of an Organization Through the Use of Analytics

An emerging area of research and practice is employing data science technologies for studying organizational dynamics, personnel behavior, and redesigning the organiza- tion to better achieve its goals. Indeed, such analytics applications are known as People Analytics. For example, analytics are used by HR departments to identify ideal candidates from the pool that submits resumes to the organization or even from broader pools such as LinkedIn. Note that with AI and analytics, managers will be able to have a larger span of control due, for example, to the advice managers and employees can get from virtual assistants. The increased span of control could result in flatter organizational structures. Also, managers’ job descriptions may have to change.

A more interesting and recent application area relates to understanding employee behavior by monitoring their movements within the organization and using that informa- tion to redesign the layout or teams to achieve better performance. A company called Humanyze (previously known as Sociometric Solutions) has badges that include a GPS and a sensor. When employees wear these badges, all of their movement is recorded. Humanyze has reportedly been able to assist companies in predicting which types of employees are likely to stay with the company or leave on the basis of their interactions with other employees. For example, those employees who stay in their own cubicles are less likely to progress up the corporate ladder than those who move about and interact with other employees extensively. Similar data collection and analysis have helped other companies determine the size of conference rooms needed or even the office layout to maximize efficiency. According to Humanyze’s Web site, one company wanted to bet- ter understand characteristics of its leaders. By analyzing the data from these badges, the company was able to recognize that the successful leaders indeed have larger net- works with which they interact, spend more time interacting with others, and are also physically active. The information gathered across team leaders was used to redesign the work space and help improve other leaders’ performance. Clearly, this may raise privacy issues, but within an organization, such studies may be acceptable. Humanyze’s Web site has several other interesting case studies that offer examples of how Big Data technolo- gies can be used to develop more efficient team structures and organizational design.

promotions targeted to the profile of each seg- ment. In addition to e-mails, special campaigns are arranged. Based on the feedback, the com- pany can plan and revise marketing strategy. SAS also helps the company to analyze the “likes” and “dislikes” of the customers. All-in- all, the intelligent systems help the company and its customers to make informed decisions.

Questions for Case 14.1

1. Why it is necessary to provide better customer experience today?

2. Why do data need sophisticated analytical tools?

3. Read the “Key benefit of SAS Marketing Automation.” Which benefits do you think are used by 1-800-Flowers.com and why?

4. Relate IBM Watson to “personalization.”

5. Relate ‘SAS Advanced Analytics’ capabilities to their use in this case.

6. ‘SAS Enterprise Miner’ is used to do data mining. Explain what is done and how.

7. SAS has a product called ‘Enterprise Guide’ that 1-800-Flowers.com uses. Find how it is used based on the tools’ capabilities.

Sources: Compiled from J. Keenan. (2018, February 13). “1-800-Flowers.com Using Technology to Win Customers’ Hearts This Valentine’s Day.” Total Retail; S. Gaudin. (2016, October 26). “1-800-Flowers Wants to Transform Its Business with A.I.” Computer World; SAS. (n.d.). “Customer Loyalty Blossoms with Analytics.” SAS Publication, sas.com/en_us/customers/1-800- flowers.html/ (accessed July 2018).

744 Part V • Caveats of Analytics and AI

Intelligent Systems’ Impact on Managers’ Activities, Performance, and Job Satisfaction

Although many jobs may be substantially enriched by intelligent technologies, other jobs may become more routine and less satisfying. Some claim that computer-based informa- tion systems in general may reduce managerial discretion in decision making and lead managers to be dissatisfied. However, studies of automated decision systems found that employees using such systems, especially those who are empowered by the systems, were more satisfied with their jobs. If using an AI system can do routine and mundane work, then it should free managers and knowledge workers to do more challenging tasks.

The most important task of managers is making decisions. Intelligent technologies can change the manner in which many decisions are made and can consequently change managers’ job responsibilities. For example, some researchers found that a decision sup- port system improved the performance of both existing and new managers as well as other employees. It helped managers gain more knowledge, experience, and expertise and consequently enhanced the quality of their decision making. Many managers report that intelligent systems have finally given them time to get out of the office and into the field. They have also found that they can spend more time planning activities instead of putting out fires because they can be alerted to potential problems well in advance thanks to intelligent system technologies (see the opening vignette, Chapter 1).

Another aspect of the managerial challenge lies in the ability of intelligent technolo- gies to support the decision-making process in general and strategic planning and control decisions in particular. Intelligent systems could change the decision-making process and even decision-making styles. For example, information gathering for decision making is completed much more quickly when algorithms are in use. Research indicates that most managers tend to work on a large number of problems simultaneously, moving from one to another as they wait for more information on their current problem. Intelligent technolo- gies tend to reduce the time required to complete tasks in the decision-making process and eliminate some of the nonproductive waiting time by providing knowledge and information.

The following are some potential impacts of intelligent system on managers’ jobs:

• Less expertise (experience) is required for making many decisions. • Faster decision making is possible because of the availability of information and the

automation of some phases in the decision-making process (see Chapters 2 and 11). • Less reliance on experts and analysts is required to provide support to top manag-

ers and executives. Today, they can decide by themselves with the help of intel- ligent systems.

• Power is being redistributed among managers. (The more information and analysis capability they possess, the more power they have.)

• Support for complex decisions makes solutions faster to develop and of better quality.

• Information needed for high-level decision making is expedited or even self-generated.

• Automation of routine decisions or phases in the decision-making process (e.g., for frontline decision making and using automated decision making) may elimi- nate some managers.

Source: Decision Support And Business Intelligence Systems, Pearson Education India, 2008.

In general, it has been found that the job of middle managers is the most likely job to be automated. Midlevel managers make fairly routine decisions, which can be fully automated. Managers at lower levels do not spend much time on decision making.

Chapter 14 • Implementation Issues: From Ethics and Privacy to Organizational and Societal Impacts 745

Instead, they supervise, train, and motivate nonmanagers. Some of their routine decisions, such as scheduling, can be automated; other decisions that involve cognitive aspects may not be automated. However, even if managers’ decisional role is completely automated, many of their other activities could not be automated or could only be partially automated.

Impact on Decision Making

Throughout the book, we illustrate how intelligent technologies improve or automate decision making. These technologies, of course, will impact managers’ job. One aspect is the impact of intelligent technologies supported by the “cloud.” An example is illustrated in Chapter 9, Figure 9.12. It illustrates the flow of data from data sources and services via an information service to analytical services for different types of decision making supported by analytics.

Uzialko (2017) describes how humans can use AI to predict and analyze the conse- quences of different potential solutions, streamlining the decision-making process. Also, by using machine learning and deep learning, more decisions can be automated.

One impact of intelligent systems is to support real-time decision making. A popu- lar tool for doing just that is SAS® Decision Manager, which is described in Technology Insights 14.1.

TECHNOLOGY INSIGHT 14.1 SAS Decision Manager

SAS Real-Time Decision Manager (RTDM) is an analytics-based integrated product that is designed to support real-time decision making, which is necessary for helping companies respond to rap- idly changing marketing, customers’ demands, technology, and other business environments.

SAS answers the following questions:

1. What does SAS RTDM do? It combines SAS analytics with business logic and contact strategies to deliver enhanced real-time recommendations and decisions to interactive customer channels, such as Web sites, call centers, point of sales (POS) locations, and automated teller machines (ATMs).

2. Why is SAS RTDM important? It helps you make smarter decisions by automating and applying analytics to the decision process during real-time customer interactions. By suc- cessfully meeting each customer’s specific needs at the right time, the right place, and in the right context, your business can become more profitable.

3. For whom is SAS® RTDM designed? It provides distinct capabilities for marketers who define communication strategies, executives who need reports on marketing ef- fectiveness, business analysts who model and predict customer behavior, and campaign managers who create target customer segments.

The following are the key benefits of RTDM:

• Makes the right decisions every time, all the time. • Realizes customer needs with the right offer, at the right time, in the right channel. • Better allocates valuable IT resources.

The key features according to SAS Inc. are:

• Real-time analytics. • Rapid decision process construction. • Enterprise data throughout. • Campaign testing. • Automated self-learning analytical process. • Connectivity.

For the details, visit “SAS Real-Time Decision Manager” and read the text there. Also you can down- load a white paper about RTDM there.

746 Part V • Caveats of Analytics and AI

Industrial Restructuring

A few authors have begun to speculate on the impact of AI, analytics, and cognitive computing on the future of industry. A few interesting resources to consult are Autor (2016), Ransbotham (2016), a special report by The Economist (Standage, 2016), and a book by Brynjolfsson and McAfee (2016). The report by The Economist is quite compre- hensive and considers many dimensions of the impact of the current developments on industry and society. The main arguments are that technology is now enabling more and more tasks that were done by humans using computers. Automating work, of course, has happened before, since the time of the Industrial Revolution. What makes the change this time around significantly more far reaching is that the technology is enabling many cognitive tasks to be done by machines. And the speed of change is so radical that the likely impact on organizations and society will be very significant and at times unpre- dictable. These authors do not agree in their predictions, of course. Let us focus first on the organizational impacts. Ransbotham (2016) argues that cognitive computing will convert many jobs done by humans to be done by computers, thus reducing costs for organizations. The quality of output may increase as well in cognitive work, which has been shown in several studies that compare a human’s performance with a machine. Everyone is aware of IBM Watson having won in Jeopardy! or Google’s system win- ning in the game of GO against human champions. But many other studies in specific domains such as speech recognition and medical image interpretation have also shown similar superiority of automated systems when the task is highly specialized yet routine or repetitive. Also, because machines tend to be available at all hours and at all loca- tions, an organization’s reach may increase, resulting in easier scaling and thus greater competition among organizations. These organizational impacts mean that yesterday’s top organizations may not remain at the top forever because cognitive computing and automation can challenge established players. This is the case in the automotive indus- try. Although traditional car companies are trying quickly to catch up, Google, Tesla, and other technology companies are disrupting industry structure by challenging the leaders of the automotive age. Analytics and AI are empowering many of these changes.

u SECTION 14.5 REVIEW QUESTIONS

1. List the impacts of intelligent systems on managerial tasks. 2. Describe new organizational units that are created because of intelligent systems. 3. Identify examples of analytics and AI applications used to redesign workspace or

team behavior.

4. How is cognitive computing affecting industry structure and competition? 5. Describe the impacts of intelligent systems on competition. 6. Discuss the impact of intelligent systems on decision making.

DisCussion Questions

1. What improvements to the decision-making process are made by SAS RTDM?

2. What SAS products are embedded or connected to RTDM? (You need to read the Web site’s details.)

3. Relate the product to product recommendation capability.

Source: SAS® Real-Time Decision Manager Make context-based marketing decisions during your real-time customer interactions. Copyright © 2018 SAS Institute Inc., Cary, NC, USA. All Rights Reserved. Used with permission.

Chapter 14 • Implementation Issues: From Ethics and Privacy to Organizational and Societal Impacts 747

14.6 IMPACTS ON JOBS AND WORK

One of the most discussed and debated topics in considering the impacts of intelligent systems is on jobs and work. There is a general agreement that:

• Intelligent systems will create many new jobs as automation always has. • There will be a need to retrain many people. • The nature of work will be changed.

The discussions, debates, and disagreements relating to the issues of when, how much, and how to deal with these phenomena occupy many researchers and are the topics of this section.

An Overview

According to Ransbotham (2016), financial advising is typically considered a knowledge- intensive task. As robot (robo) advisors provide personalized support for individuals, the costs of such services go down. This leads more people to demand such services, eventu- ally freeing more humans to address advanced financial issues. Robo advisors may also cause some people to lose their jobs.

Some authors argue that the automation segment, which is related to cognitive computing and AI, will accelerate what is called polarization of the labor market in the future. This entails significant job growth in the top and bottom tiers of the labor market but losses in the middle. Jobs requiring low but specialized skills, such as personal care, are continuing to grow. Similarly, jobs that require very high skill, such as graphics design work, and so on, are also growing. But jobs that require “middle skills” such as special- ized knowledge that is applied over and over with some adaptation, are at the greatest risk of disappearing. Sometimes technology disintermediates itself! For example, IBM Watson Analytics now includes querying capabilities to begin asking questions that an intelligent system professional previously asked and, obviously, providing answers. Other analytics-as-a service offerings with similar services may result in a need for fewer people to be proficient at using analytics software.

A report by The Economist notes that even if AI does not replace workers directly, it will certainly require employees to acquire new skills to keep their jobs. Market disrup- tion is always uncomfortable. The next few years will provide excellent opportunities for intelligent technology professionals to shape the future.

Are Intelligent Systems Going to Take Jobs—My Job?

Tesla’s Elon Musk envisions AI-based autonomous driving trucks all over the world within 10 years. There will be convoys of such trucks, each of which will follow a lead truck. Trucks will be electrical, economical, and pollution free. In addition, there will be fewer accidents—sounds great! But what about thousands of drivers who will lose their jobs? What about many thousands of employees in truck stops who will lose their jobs as well? The same scenario could happen in many other industries. Amazon has opened its first Go, a cashierless physical store. They plan 3,000 more in a few years. The post office in some countries already distributes mail using autonomous vehicles. In short, there is a chance for massive unemployment.

Example: Pilots at FedEx

FedEx has a fleet of close to 1,000 airplanes flying globally. According to Frank Tode, editor and publisher of The Robot Report, FedEx hopes that around 2020 the company will have one global pilot center with three or four pilots who will operate the entire FedEx fleet.

748 Part V • Caveats of Analytics and AI

While it may be sometime before FedEx will have pilotless airplanes and schools will have no human teachers, some jobs, according to Sherman (2015), have already been taken by robots. They include:

• Online marketers. Using NLP, companies are automatically developing marketing ads and e-mails that influence people to buy (robo marketers). These are based on a dialog with potential buyers and on an automatic database search of historical cases. “Who needs an on-

line marketer that may have inferior, biased, or incomplete knowledge?”

• Financial analysts and advisors. As was described in Chapter 12, robo advisors are all over the scene. Equipped with the ability to deal with Big Data in real time and conduct predic- tive analysis in seconds, these programs are liked by investors who pay about one-tenth of what human advisors charge. Furthermore, robo advisors can personalize recommendations.

Application Case 14.2 White-Collar Jobs That Robots Have Already Taken

Foxcom, an iPhone manufacturer in Taiwan, had planned to replace almost all of its em- ployees (60,000) in Taiwan with robots (Botton, 2016). The company already produced 10,000 robots for this purpose.

INTELLIGENT SYSTEMS MAY CREATE MASSIVE JOB LOSSES The debate regarding tech- nology taking jobs has been going on since the beginning of the industrial revolution. The issue regarding intelligent systems is strongly debated now due to the following:

• They are moving very fast. • They may take a large variety of jobs, including many white-collar and nonphysi-

cal jobs. • Their comparative advantage over manual labor is very large and growing rapidly

(see Figure 2.2 in Chapter 2). • They are already taking some professional jobs from financial advisors, paralegals,

and medical specialists. • The capabilities of AI are growing rapidly. • In Russia, robots are already teaching mathematics in schools (some do a better

job than humans). Just think about what could happen to the teaching profession.

AI Puts Many Jobs at Risk

For the potential impact of AI on jobs, see Dormehl (2017), who explores the possibility of creative intelligent machines. For example, McKinsey’s study estimates that AI is poised to take over 30 percent of all bank jobs in the near future. The study also predicts that robots will take 800 million jobs worldwide by 2030 (Information Management News, 2017).

To research the potential danger of job loss, McKinsey & Company divided jobs into 2,000 distinct work activities, such as greeting customers and answering questions about products, which retail salespeople do. Its researchers (see Chui et al., 2015) found that 45 percent of all 2,000 activities could be economically and physically automated. The activi- ties include physical, cognitive, and social types.

While autonomous vehicles are not taking jobs, yet, they will take jobs from taxi drivers, Uber, and similar companies’ drivers. Also, bus drivers may lose their jobs. Other jobs that have already been replaced by intelligent systems are listed in Application Case 14.2.

Chapter 14 • Implementation Issues: From Ethics and Privacy to Organizational and Societal Impacts 749

Let us look at some other studies. A 2016 study done in the United Kingdom pre- dicted that robots will take 50 percent of all jobs by 2026. Egan (2015) reports that ro- bots already threaten the following jobs: marketers, toll booth operators and cashiers, customer service, financial brokers, journalists, lawyers, and phone workers. Note that automation may affect portions of almost all jobs to a greater or lesser degree. Experts estimate that about 80 percent of IT jobs may be eliminated by AI.

According to Manyika et al. (2017), automation is spreading because “robots are also increasingly capable of accomplishing activities that include cognitive capabilities once considered too difficult to automate successfully, such as making tacit judgments, sensing emotion, or even driving.”

Given all this, you may wonder whether your job is at risk.

Which Jobs Are Most in Danger? Which Ones Are Safe?

If want to know about your job, it obviously depends on the type of job you are holding. Oxford University in the United Kingdom looked at 700 jobs and ranked them from zero (no risk of automation) to 1 (very high risk of automation). Straus (2014) provided a list of the top 100 most at-risk jobs (all above 0.95) and the 100 jobs with the lowest risk (with 0.02 or less). The top 10 “safe” and the 10 at risk are listed in Table 14.1.

A 2017 study conducted by the Bank of England found that almost half of the U.K. jobs (15 million out of 33.7 million) are at risk of loss within 20 years. Creative robots are the great- est threat because they can learn and increase their capabilities. While in the past, automation may not have decreased the total number of jobs, this time the situation may be different.

A side effect of this situation may be that workers will have less income while the owners of robots will have a larger income. (This is why Bill Gates suggested taxing the robots and their owners.)

• Anesthesiologists, diagnosticians, and sur- geons. The medical field seems to be immune from AI. This is not the case. Expert systems for diagnosis have been in place for about 40 years. The FDA has already approved the J&J Sedasys system for delivery of low-level anesthesia in sur- geries, such as colonoscopies. IBM’s Watson has demonstrated a far more accurate diagnosis in lung disease cases than humans (90% vs. 50%). Finally, surgeons already use automated ma- chines in some invasive procedures.

• Financial and sports reporters. These jobs involve gathering information, interviewing people, answering questions, analyzing the material, and writing reports. The Associated Press (AP) has experimented with AI machines since 2014. Results so far are virtually error and bias free (and no fake news!).

Palmer (2017) reported an additional five jobs in danger, including middle management, com- modity salespeople, report writers, accountants and bookkeepers, and some types of doctors.

McFarland (2017b) lists as high-risk jobs cashiers, toll booth operators, fast-food employees,

and drivers. Low-risk jobs include nurses, doctors, dentists, youth sport coaches, and social workers.

Questions for Case 14.2

1. Watch the 4:22 min. video about an interview with Palmer, at linkedin.com/pulse/5-jobs- robots-take-first-shelly-palmer/. Discuss some of the assertions made regarding doctors.

2. Discuss the possibility of your checkup by a robot-diagnostician. How would you feel?

3. With the bombardment of fake news and their biased creators, it may be wise to replace all of them by intelligent machines. Discuss such a possibility.

4. You are a defendant in a crime you did not com- mit. Would you prefer a traditional lawyer or one equipped with an AI e-discovery machine? Why?

Sources: Compiled from E. Sherman. (2015, February 25). “5 White-Collar Jobs Robots Already Have Taken.” Fortune.com. fortune.com/2015/02/25/5-jobs-that-robots-already-are- taking (accessed April 2018); S. Palmer. (2017, February 26). “The 5 Jobs Robots Will Take First.” Shelly Palmer.

750 Part V • Caveats of Analytics and AI

TABLE 14.1 Ten Top Safe and at Risk Occupations

Source: Based on Straus (2014) Straus, R.R. “Will You Be Replaced by a Robot? We Reveal the 100 Occupations Judged Most and Least at Risk of Automation.” ThisisMoney.com, May 31, 2014. thisismoney.co.uk/money/news/article-2642880/ Table-700-jobs-reveals-professions-likely-replaced-robots.html

Probability of Job Loss

Low-Risk Jobs

0.0036 First-Line supervisors of firefighting and prevention workers

0.0036 Oral and maxillofacial surgeons

0.0035 Healthcare social workers

0.0035 Orthotists and prosthetists

0.0033 Audiologists

0.0031 Mental health and substance abuse social workers

0.0030 Emergency management directors

0.0030 First-Line supervisors of mechanics, installers, and repairers

0.0028 Recreational therapists

High-Risk Jobs

0.99 Telemarketers

0.99 Title examiners, abstractors, and searchers

0.99 Sewers, hand

0.99 Mathematical technicians

0.99 Insurance underwriters

0.99 Watch repairer

0.99 Cargo and freight agents

0.99 Tax preparers

0.99 Photographic process workers and processing machine operators

0.99 New account clerks

SOME MORE JOB LOSSES OBSERVATIONS • Kelly (2018) predicts that robots could eliminate many Las Vegas jobs. And indeed,

in many casinos worldwide, you can play several traditional games on machines. • People with doctoral degrees have a 13 percent chance of being replaced by robots

and AI versus 74 percent for those with only a high school education (Kelly, 2018). • Women will lose more jobs to automation than men (Krauth, 2018).

Intelligent Systems May Actually Add Jobs

Despite the fear, uncertainty, and panic related to job losses, many reports contradict this. Here are some examples: de Vos (2018) reported that AI will create 2 .3 million jobs in 2020 while eliminating 1 .8 million. Also, one needs to consider the great benefits of AI and the fact that human and machine intelligence will complement each other in many jobs. Also, AI will increase international trade, adding more jobs. de Vos also cites studies that show the creation of jobs due to equipment maintenance and service that cannot be automated. The following are predictions on both sides of the issue:

• A PricewaterhouseCoopers (PwC) study forecast that robots will bolster U.K. eco- nomic growth. So, even though robots could destroy about 7 million jobs in the United Kingdom, they will create at least 7 million new jobs and probably more over 20 years (Burden, 2018).

Chapter 14 • Implementation Issues: From Ethics and Privacy to Organizational and Societal Impacts 751

• IBM’s new deep learning service may help save IT jobs. • There is a shortage of millions of skilled workers (e.g., about 50,000 truck drivers in

the United States), so automation will reduce millions of unfilled positions. • Korolov (2016) claims that there is plenty of work, especially for people who keep

up with technology and broaden their skills. • Gartner Inc. predicts that by 2020, AI will create more jobs than it eliminates Singh,

(2017b). • Wilson et al. (2017) report on new categories of human jobs that have been created

by AI. • Some believe that there will be a total of increase in jobs due to AI-induced

innovations. • It was estimated that in 2018 there would be over 490,000 jobs open for data scien-

tists, but only 200,000 scientists will be available. However, in the long run, AI and machine learning may replace most data scientists (Perez, 2017).

• Violino (2018) contradicts those who claim that there is a huge fear among employ- ees regarding job loss, saying that most workers see robots as an aid to their jobs. See also Leggatt (2017).

Note: When this book went to press, there was a shortage of IT employees (several million in the United States). Automation can alleviate this shortage. Note that a study reported by Weldon (2017b) showed that most workers actually welcome the impact on jobs by AI and automation. As a final note, Guha (2017) provides a view of work and AI as a vision of “despair, hope, and liberation.” He concludes that AI can liberate work—it is a historical opportunity.

Jobs and the Nature of Work Will Change

While you may not lose your job, intelligent applications may change it. One aspect of this change is that low-skill jobs will be taken by machines, but high-skill jobs may not. Therefore, jobs may be redesigned either to be low skilled in order to be automated, or to be high skilled so that they will be executed exclusively by humans. In addition, there will be many jobs where people and machines will work together as a team.

Changes in jobs and business processes will impact training, innovation, wages, and the nature of work itself. Manyika (2017) and Manyika et al. (2017) of McKinsey & Company analyzed the shifts that can be fundamental, and arrived at the following conclusions:

• Many activities done by humans will have the potential to be automated. • Productivity growth from robotics, AI, and machine learning will be tripled com-

pared to pre-2015. • AI will create many new jobs paying high salaries. • Since more than half the world is still offline, the changes will not be too rapid.

Example: Skills of Data Scientists Will Change

According to Thusoo (2017) of the McKinsey Global Institutes study group, there will be a shortage of 250,000 data scientists by 2024. There will be a need to retrain or train scientists so they can deal with intelligent technologies and the changes in data science and in solv- ing related real-world problems. Thus, proper education must evolve. The job requirements of data scientists are already changing. The scientists will need to know how to apply ma- chine learning and intelligent technologies to build IoT and other useful systems. New al- gorithms improve operations and security, and data platforms are changing to fit new jobs.

Snyder (2017) found that 85 percent of executives know that intelligent technolo- gies will impact their workforce within five years, and 79 percent expect the current

752 Part V • Caveats of Analytics and AI

DEALING WITH THE CHANGES IN JOBS AND THE NATURE OF WORK Manyika (2017) made the following suggestions for policymakers:

1. Use learning and education to facilitate the change. 2. Involve the private sector in enhancing training and retraining. 3. Have governments provide incentives to the private sector so employees can invest

in improved human capital. 4. Encourage private and public sectors to create appropriate digital infrastructure. 5. Innovative income and wage schemes need to be developed. 6. Carefully plan the transition to the new work. Deal properly with displaced

employees. 7. Properly handle new technology-enabled technologies. 8. Focus on new job creation, particularly digital jobs. 9. Properly capture the productivity increase opportunities.

Baird et al. (2017) of McKinsey & Company provide a video interview with industry ex- perts discussing how to deal with the changing nature of work. Another exploration of the nature of work in the era of intelligent systems is provided by Crespo (2017). Chui et al. (2015) researched the impact of automation on redefining jobs and business pro- cesses, including the impact on wages, and the future of creativity. Finally, West (2018) provides a comprehensive study on the future of work as it is influenced by robotics and AI-driven automation.

Conclusion: Let’s Be Optimistic!

Assuming that the disasters will not occur, then, as in the past, concerns about technol- ogy replacing many human jobs and reducing wages are hopefully exaggerated. Instead, intelligent technologies will clearly contribute to shorter work time for humans. Today, most people work long hours just for survival.

u SECTION 14.6 REVIEW QUESTIONS

1. Summarize the arguments of why intelligent systems will take away many jobs. 2. Discuss why job losses may not be catastrophic. 3. How safe is your job? Be specific.

• Digital capabilities need to come before AI. • Machine learning is powerful, but it is not the solution to all problems. • Do not put technology teams solely in charge of intelligent technologies. • Adding a business partner may help with AI-based projects. • Prioritize a portfolio approach to AI initiatives. • The biggest challenges will be people and business processes. • Not every business is using intelligent systems, but almost all those that use them

increase income and profit. • Top leadership support is necessary for a transformation to AI.

skill sets to be restructured. They also expect 79 percent productivity improvement. Employees  fear that intelligent systems will take over some of their activities, but they hope that intelligent systems will also help with their work.

TIPS FOR SUCCESS A McKinsey study of 3,000 executives (Bughin, McCarthy, and Chui, 2017) reports the following success tips for implementing AI provided by the executives:

Chapter 14 • Implementation Issues: From Ethics and Privacy to Organizational and Societal Impacts 753

4. How may intelligent systems change jobs? 5. In what ways may work be changed? 6. Discuss some measures to deal with the changes brought by intelligent systems. 7. One of the areas of potential job loss is due to autonomous vehicles. Discuss the

logic of this.

14.7 POTENTIAL DANGERS OF ROBOTS, AI, AND ANALYTICAL MODELING

During the period 2016–2018, we witnessed a heated debate regarding the future of AI and particularly robots. Dickson (2017) called the optimistic approach Utopia and the pessimistic one Dystopia. The debate began with the industrial revolution regarding au- tomation, and it has accelerated because of the rapid technological innovations of AI. In Section 14.5, we presented one aspect of this debate, the impact on jobs. In the center of the debate is the prediction of when AI’s capabilities to reason and make decisions will become similar or even superior to that of people. Furthermore, will such a development be beneficial or dangerous to society?

Position of AI Dystopia

The camp that supports this prediction includes well-known tech executives. Here are three of them:

• Elon Musk: “We need to be super careful with AI. Potentially more dangerous than nukes.” (See the 10 min . video at youtube.com/watch?v=SYqCbJ0AqR4). Musk predicts that World War III will start because of AI. “Robots will kill us all, one day,” he said in his several presentations.

• Bill Gates: “I am in the camp that is concerned about super intelligence. Musk and some others are on this and I don’t understand why some people are not con- cerned.” (Comments made on TV and interviews, several times). He also suggested taxing the manufacturers and users of robots and other AI machines.

• Stephen Hawking: The late scientist stated, “The development of full artificial intelligence could spell the end of the human race.”

Many people are afraid of AI because they believe that computers will become smarter than we are. See Bostrom’s video of his famous TED presentation at youtube.com/ watch?v=MnT1xgZgkpk. See also Maguire (2017) for a discussion regarding learning ro- bots and the risk of rebelling robots. For how robots can learn motor skills through trial and error, see the video at youtube.com/watch?v=JeVppkoloXs/. For more, see Pham (2018).

The AI Utopia’s Position

A good place to begin for information on this position is to watch the 26 min. documen- tary video on the future of AI at youtube.com/watch?v=UzT3Tkwx17A. This video concentrates on the contribution of AI to the quality of life. One example is crime fighting in Santa Cruz, California, where AI was able to predict where and when crimes will occur. Following the predictions, the police department has been planning its work strategies. The result is a 20 percent reduction in crime.

A second example is the prediction of the probability that a certain song will be a hit. The prediction helps both artists and managers to plan their activities. Great success has been made. In the future, AI is predicted to compose top songs.

Finally, there is a story about dating. The capabilities of AI enabled a scientist to find a perfect match in a population of 30,000 potential candidates.

754 Part V • Caveats of Analytics and AI

A basic argument of the Utopianists expressed in interviews, TV lectures, and more, is that AI will support humans and enable innovations. AI also will partner with humans. The Utopians believe that as AI expands, humans will become more productive and will have time to do more innovative tasks. At the same time, more tasks will be fully auto- mated. Prices of products and services will drop and the quality of life will increase.

At one point, we may achieve a fully automated and self-sustaining economy. Ultimately, people will not have to work at all to make a living.

A leading proponent of AI benefits is Mark Zuckerberg of Facebook. He is in a heated debate with Elon Musk (CEO of Tesla Corp), the unofficial leader of the Dystopia camp of believers. Zuckerberg criticized those that believe that AI will cause “doomsday scenarios” (see the next section). Musk claimed that Zuckerberg has a “limited under- standing” of AI, and Zuckerberg answered by referring to his paper on AI that won an award at the “top computer vision conference.” For details, see Vanian (2017).

SOME ISSUES RELATED TO THE UTOPIA Several issues are related to the Utopianists’ po- sition. Here are three examples:

1. AI will be so great that people will have a problem of what to do with their free time. If you have not yet seen Disney’s Wall-E movie, go and see it. It shows how humans are served by robots. Dennis Hassabis, a strong proponent of Utopia (from Deep Mind, an AI company), believes that AI will one day help people have a better life by understanding what makes humans unique, what the mysteries of the mind are, and how to enjoy creativity.

2. The road to AI Utopia could be rocky, for example, there will be impacts on jobs and work. It will take time to stabilize and adjust work and life of living with robots, chatbots, and other AI applications.

3. One day we will not drive anymore and there may not be human financial advisors; everything will be different, and the changes may be rapid and turbulent and we may even face disasters, as projected by the Dystopia camp.

The Open AI Project and the Friendly AI

To prepare against the unintended action of robotics and AI, Elon Musk and others have created Open AI, a nonprofit organization. With the unintended potential danger in mind, Musk and others created a nonprofit AI research company endowed with $1 billion. The major objective is to enact the path to safe artificial general intelligence (AGI). As you recall from Chapter 1, AGI is not here yet, but it is coming.

The plan of Open AI is to build safe AGI and ensure that its benefits will be evenly distributed. The research results are published in top journals. In addition, Open AI cre- ates open source software tools. The organization has a blog and it disseminates impor- tant AI news. For details, see openai.com.

THE FRIENDLY AI Eliezer Yudkowsky, a cofounder of the Machine Intelligence Research Institute, developed the idea of friendly AI, according to which AI machines should be designed so that they will benefit humans rather than harm them (i.e., use a system of checks and balances in designing the AI capabilities). For details, see Sherman (2018), and view a fascinating 1:29:55 min . video by Yudkowsky (2016) at youtube.com/ watch?v=EUjc1WuyPT8.

CONCLUSION It is difficult to know what will happen in the future. But some actions are already being taken to prevent a disaster. For example, several major companies have declared that they will not produce or support killer robots.

Chapter 14 • Implementation Issues: From Ethics and Privacy to Organizational and Societal Impacts 755

The O’Neil Claim of Potential Analytics’ Dangers

Managers and data science professionals should be aware of the social and long-term ef- fects of mathematical models and algorithms. Cathy O’Neil, a Harvard PhD in mathemat- ics who worked in finance and the data science industry, expressed her experiences and observations in the popular book Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy. We suggest you read the book, or at least visit the author’s blog site at mathbabe.org/. The blog site highlights social issues related to ana- lytics. A good summary/review of the book is available at knowledge.wharton.upenn. edu/article/rogue-algorithms-dark-side-big-data/.

In her book, O’Neil (2016) argues that models must satisfy three conditions. First, they must be transparent. That is, if the model is not understandable, its application can lead to unintended consequences.

Second, the model must have clear quantifiable objectives. For example, the cel- ebrated application of analytics in the book and movie Moneyball includes a model that was aimed at increasing the number of financial wins. And the proposed input measures were well understandable. Rather than using the more commonly reported measure “run base in” (RBI), the analyst in Moneyball proposed and used on-base percentage and other measures (which were also easily calculated and understood by anyone with basic math skills). On the other hand, models built to assess the risk of mortgage-backed securities when no one fully understood the underlying assumptions of collateralized securities, but financial traders were trading, have been blamed for leading the financial crisis of 2008.

The third requirement is that the models must have a self-correcting mechanism and a process in place so that they are audited regularly and new inputs and outputs are constantly being considered. This third issue is particularly critical in applying models in social settings. Otherwise, the models perpetuate the faulty assumptions inherent in the initial modeling stage. O’Neil discusses several situations where such is the case. For example, she describes the models built in the United States to identify underperform- ing teachers and reward better teachers. Some of these models utilized the test scores of the pupils to assess the teachers. O’Neil cited several examples where the models were used to fire “underperforming” teachers even though those teachers were loved by the students and parents. Similarly, models are used to optimize the scheduling of workers in many organizations. These schedules may have been developed to meet seasonal and daily demand variations, but the models do not take into account the deleterious impacts of such variability in schedules on the families of these usually lower-income workers. Other such examples include credit score assessment models that are based on historical profiles and thus may negatively impact minorities. Without mechanisms to audit such models and their unintended effects, they can do more harm than good in the long term. So, model builders need to consider such concerns.

Note: In May 2018, General Data Protection Regulation (GDPR) became effective in the European Union. It includes the need to explain data. According to Civin (2018), an explainable AI could reduce the impact of biased algorithms.

A comment: There is evidence that in some cases O’Neil’s claims are valid, and therefore model builders and implementers must pay attention to the issues. However, in general, analytics are properly designed and bring considerable benefits to society. Furthermore, analytical models increase the competitiveness of companies and countries, creating many highly paid jobs. In many cases, companies have social responsibility policies that minimize biases and inequality. Finally, as Weldon (2017a) observed, algo- rithms and AI can be seen as great equalizers in bringing services that were traditionally reserved for a privileged few, to everyone.

756 Part V • Caveats of Analytics and AI

u SECTION 14.7 REVIEW QUESTIONS

1. Summarize the major arguments of the Utopia camp. 2. Summarize the major arguments of the Dystopia camp. 3. What is the friendly AI? 4. What is Open AI? Relate it to the dystopia vision. 5. What are the potential risks in using modeling and analytics?

14.8 RELEVANT TECHNOLOGY TRENDS

As we near the last section of this book that discusses some aspects of the future of intelligent systems, it is worthwhile to describe some of the technology trends that will shape this future. Unfortunately, there are hundreds of technology trends relevant to the content of this book. The reason is that there are hundreds of variations of analytics, Big Data tools, AI, machine learning, IoT robotics, and other intelligent systems. Therefore, we provide here only a sample of technology trends. We divide this section into the fol- lowing subsections:

• Gartner’s 2018 and 2019 lists. • List of technology trends in intelligent systems. • Ambient computing.

Gartner’s Top Strategic Technology Trends for 2018 and 2019

Gartner Inc. is a top technology research organization and consultant as well as an or- ganizer of an annual technology symposium attended by over 23,000 people (Gartner Symposium IT expo). It provides an annual prediction of the technologies that it thinks will impact most organizations. The 2018 and 2019 lists of trends includes 10 items each, most of which relate directly to the content of our book.

The summary of the 2018 list is shown in Figure 14.3. It was extracted from Gartner’s press release of October 4, 2017, which is available at gartner.com/newsroom/id/3812063. The essentials are provided in a video (5:36 min.) at youtube.com/watch?v=TPbKyD2bAR4.

GARTNER’S 2018 AND 2019 LISTS The following is extracted from gartner.com/news- room/id/3812063, for 2018, and from Weldon (2018), for 2019.

1. AI Foundation and Development. Advanced AI systems that support decision making, some of which are autonomous, and other AI systems are developed in con- junction with analytics and data science.

2. Intelligent Apps and Analytics. Almost all IT systems will include AI in the next few years. See[gartner.com/smarterwithgartner/the-cios-journey-to-artificial- intelligence/].

3. Intelligent and Autonomous Things. Utilizing the IoT capabilities, there will be an explosion of autonomous vehicles and a significant increase of other intelligent things (e.g., smart homes and factories where robots are assembling robots).

4. Digital Twin. A digital twin, see [gartner.com/smarterwithgartner/prepare- for-the-impact-of-digital-twins/], refers to digital representations of real-world ob- jects and systems. This includes mainly IoT systems with 20 billion connected things in two to three years.

5. Empowered Cloud (Cloud to the Edge). In Edge computing, information collection, processing, and delivery are conducted closer to the sources of the information.

Chapter 14 • Implementation Issues: From Ethics and Privacy to Organizational and Societal Impacts 757

6. Conversational Human-Machine Platforms. These platforms already facilitate natural language interactions, resulting in improved collaboration. These include smart collaborative spaces..

7. Immersive Experience. These systems change the manner in which people can see and perceive the world (e.g., augmented reality). See [gartner.com/smarterwithgartner/ transform-business-outcomes-with-immersive-technology/].

8. Blockchain. Blockchain technologies [gartner.com/smarterwithgartner/are- you-ready-for-blockchain-infographic/] offer a radical platform for increased se- curity and trust, significantly improving business transactions.

9. Augmented Analytics. Using machine learning enables this technology to focus on transformation of analytics, so it will be better shared and consumed. This will facilitate data preparation management and analysis to improve decision support.

10. Others. These include smart collaboration space, Quantum computing, digital and ethical privacy, and adopting risks and trust.

Other Predictions Regarding Technology Trends

• The IEEE computer society also has 10 top predictions for 2018. computer.org/ web/pressroom/top-technology-trends-2018. The list includes deep learning, industrial IoT, robotics, assisted transportation, augmented (assisted) reality, block- chain, and digital currencies.

• Newman (2018) provides a list of 18 tech trends at CES 2018. These are related to displays at CES.

• The potential business application and value for several analytics and AI technolo- gies based on studies of 400 real-world cases done at McKinsey & Company is available as interactive data visualization at mckinsey.com/featured-insights/ artificial-intelligence/visualizing-the-uses-and-potential-impact-of-ai-and- other-analytics/ (posted April 2018).

FIGURE 14.3 Predict the future of AI (Drawn by E.Turban)

Today Narrow AI (Chapter 2)

Increase capabilities with time, but no match to human

intelligence

In 20–25 years Artificial General

Intelligence (Chapter 2)

Autonomous systems all over; limited reasoning

capabilities; adopt to changes in the

environment; can self-expand tasks;

can reason, innovate

In Distant Future Super AI

As intelligent as human and even

more in some cases. Major support to

research, innovation and learning

Increasingly perform routine tasks, some with human. Provide speed, quality and advice. Cut cost

Autonomous vehicles are all around. Robots

assume more tasks. People have more

time. Compete with humans

Can be dangerous if not controlled. Can significantly improve

our quality of life

Possible impacts

758 Part V • Caveats of Analytics and AI

• Top 10 trends for analytics in 2018 are provided by Smith (2018). The list is fairly technical in nature. It includes “Data Gravity will accelerate to the cloud,” “Insight- as-a-service will rise,” and “End-to-end cloud analytics will emerge.”

• Top 10 AI technology trends for 2018 as envisioned by Rao et al. (2017) include “Deep reinforcement learning: interacting with the environment to solve business problems” and “Explainable AI: understanding the black box.”

• For seven data and analytical trends, see datameer.com/blog/seven-data- analytics-trends-2018/.

• Computers will learn to think and think to learn. • Robots will replace humans in more nonphysical and cognitive roles. • Intelligent augmentation is part of the narrow AI (Chapter 1) and will continue to

control new AI applications. • Edge computing was cited by Gartner, but it has much more value that may not

be related to the “cloud.” The technology will have a major impact on the future of data centers. For details, see Sykes (2018a). Note that most of the new capa- bilities for the “cloud” exist in the use of the “Edge.” For further information, visit Wikipedia. Edge AI enhancements will excel in supporting machine learning and augmented reality.

Sommer (2017) lists the following:

• Data literacy will spread both in organizations and in society. • Information points will be connected via hybrid multi-cloud systems. • The mystery of rural networks will be exposed by deep learning theory. • Self-service systems will use data catalogs as their frontier. • Need to focus on Application Programming Interfaces (APIs). • Analytics become conversational (e.g., via chatbots). • Analytics will include immersive capabilities. • Using augmented intelligence users will be turned to participants. • For 11 top trends that drove business intelligence in 2018, see Sommer (2017). • For six data analytics trends in 2018, see Olavsrud (2018). • For robotics trends in 2018, see Chapman (2018). • For 10 predictions of intelligent systems, see Press (2017).

Summary: Impact on AI and Analytics

Now that you have seen the many technologies trends for the future, you may also want to see when they will impact AI. Figure 14.3 illustrates the long-term projection of AI. The future is divided into three sections: today, in about 20 years, and in a distant future.

The future of BI and analytics is illustrated in Figure 14.4. Some additional pre- dictions are intelligent analytics, insight-as-a-service, and data cataloging. Finally, we describe one technology in more detail. It may impact both analytics and AI.

Ambient Computing (Intelligence)

Closely related to the IoT, chatbots, smart homes, analytics, sensors and “things” are included in the concept of ambient computing (or paradigm computing). It has several definitions, but essentially it refers to electronic environments (e.g., network devices such as sensors) that are sensitive and responsive to people and their envi- ronments. So ambient devices can support people in whatever task they are doing. Once sensing their surroundings, the devices provide different input/output methods that depend on the configuration of situations (e.g., what people are doing at a given

Chapter 14 • Implementation Issues: From Ethics and Privacy to Organizational and Societal Impacts 759

time). In summary, everything in our life will be computerized and intelligent. The concept is based on previous research in the areas of pervasive computing, human– machine interaction, context awareness, profiling, personalization, and interaction de- sign. For details, see en.wikipedia.org/wiki/Ambient_intelligence and Charara’s (2018) guide.

POTENTIAL BENEFITS OF AMBIENT INTELLIGENCE While the concept is mostly futuristic, its characteristics and benefits are already envisioned. The networked devices can:

• Recognize individuals and other “things” and their context at any given time and place.

• Integrate into the environment and existing systems. • Anticipate people’s desires and needs without asking (e.g., context awareness). • Deliver targeted services based on people’s needs. • Be flexible (i.e., can change their actions in response to people’s needs or

activities). • Be invisible.

FIGURE 14.4 Future of Analytics Source: “Analytics and BI Trends”, Datapine, in Top 10 Analytics and Business Intelligence Trends for 2018, Business Intelligence, Dec 13th 2017, © 2017, Used with permission.

Top 10 Business Intelligence Trends for 2018

Data Governance

Natural Language

Processing

Multi-Cloud Strategy

Predictive and Prescriptive Analytics

Tools Security

Collaborative Business

Intelligence

Embedded Business

Intelligence

Data Quality Management

Chief Data Officer

Artificial Intelligence

Many of the devices and services described in this book already exhibit some of the capabilities of ambient computing. Amazon’s Alexa is probably currently the closest to the ambient concept. For details, see Kovach (2018). For more on ambient computing and its relationship to IoT and smart cities, see Konomi and Roussos (2016).

u SECTION 14.8 REVIEW QUESTIONS

1. Identify three of the Gartner 10 that are mostly related to analytics and data science. 2. Identify three of the Gartner 10 that are most related to AI and machine learning. 3. Identify three of the Gartner 10 that are most related to IoT, sensors, and connectivity.

760 Part V • Caveats of Analytics and AI

4. Identify three technologies related to analytics from the other predictions list and explore them in more detail. Write a report.

5. Identify three data science–related technologies from the long list and explore them in more detail. Write a report.

6. Identify three AI-related technologies from the long list and explore them in more detail. Write a report.

7. Describe ambient computing and its potential contribution to intelligent systems.

14.9 FUTURE OF INTELLIGENT SYSTEMS

There is a general agreement among AI experts that AI is going to change everything in our world for the better (e.g., see Lev-Ram [2017] and Violino [2017]). However, there are disagreements on when such changes will occur and what their impact is going to be. AI research is accelerating due to improvements in different related computer technologies (e.g., chips, IoT), improvements in intelligent methodologies and tools, the increased activities in high-tech companies that are striving to gain leadership in certain intelligent systems areas and firms that are investing billions of dollars in AI, the development of AI tools and methodologies, and much more. In this section, we first provide a presentation of what some major corporations are doing in the intelligent technologies field.

What Are the Major U.S. High-Tech Companies Doing in the Intelligent Technologies Field?

One way to predict the future of AI is to look at what the major companies are currently doing.

GOOGLE (ALPHABET) Google uses NLP in its Google Translate as well as in its search processes. It uses neural networks in its immersed databases (for pattern recogni- tion) and for making decisions on them. In addition, Google uses other machine- learning algorithms for personalization advertising decisions. Google Assistant and Home are two applied projects that attracted considerable attention in CES 2018. Google Assistant is trying to dethrone Alexa. In addition, Google is most active in the autonomous vehicles field. Google purchased several AI companies and is conduct- ing extensive research in the field. Google has a special team that attempts to provide Google AI speech dialog with a personality (see Eadicicco, 2017). Google DeepMind’s AlphaGo is the machine that beat the game Go champions. Google is using machine learning for managing its huge databases and search strategies. Finally, Google is teaching its AI machines how people behave (e.g., cook, huy) by showing them film clips (see Gershgorn, 2017).

APPLE Apple is known to secretly be working on several AI projects. The most known is its Siri chatbot, which is embedded in several of its products (e.g., iPhone). In 2016, Apple acquired a machine-learning company, Turi. While lagging behind Google, Amazon, and Microsoft, Apple is rapidly closing the gap, using acquisitions and extensive research and development. Apple acquired companies in speech recognition (Vocal!), image recogni- tion (Perception), and facial expression recognition (Emotion). Thus, Apple is becoming a leader in AI. With several hundred millions of Siri users and new acquisitions in AI, Apple is charging forward rapidly.

FACEBOOK Mark Zuckerberg, Facebook’s CEO, is a major believer in the future of AI. In addition to his personal investments in AI, he hired Yann LeCun, a deep-learning pioneer,

Chapter 14 • Implementation Issues: From Ethics and Privacy to Organizational and Societal Impacts 761

to lead AI research in the company. LeCun created a special Facebook unit that identifies important AI developments and incorporates them into Facebook’s products. Facebook invested billions of dollars in AI. With Facebook, AI goes mainstream. With its over 2 bil- lion users, Facebook is spreading its AI applications globally.

MICROSOFT Microsoft is very active in all AI technology research. In 2017, it acquired Maluuba, a start-up that specializes in deep learning and NLP. Some believe that this acquisition will help Microsoft outperform both Facebook and Google in the areas of speech and image recognition. Maluuba excels in reading and comprehending text with near human capabilities in its virtual personal assistant, Cortana. This assistant helps people deal with e-mail and messaging difficulties. The AI will examine the content of messages and any stored documents and advice for what actions to take. For a compre- hensive video about AI today and tomorrow by Stanford University, watch a 74 min. seminar at youtube.com/watch?v=wJqf17bZQsY.

IBM IBM entered robotics as early as 1973. By 1980, it had developed the QS-1; by 1977, it had developed Deep Blue; and by 2014, a mature IBM Watson entered the scene. IBM is also known for its artificial brain project. (For Blue Brain, see artificialbrains.com/ blue-brain-project.) IBM is also known for its Deep QA project.

IBM is very active in AI research, especially in the area of cognitive computing; see Chapter 6 and research.ibm.com/ai/. IBM Watson was developed in collaboration with MIT AI labs.

Some other current projects focus on distributed deep learning software, creation of music and movie trailers by machines, gesture recognition, combining AI and IoT (e.g., embodied cognition), and medical applications supported by Watson (cognitive care, e.g., cancer detection, mental health care, and visually impaired people). IBM Watson is already considered the strongest applied brand of AI. One billion users were expected to use it in 2018, gaining substantial benefits from its applications.

AI Research Activities in China

AI research is done in many countries, notably Germany, Japan, France, the United Kingdom, and India. But most research outside the United States is done in China. China plans to be the world leader in AI, and its government is strongly supporting the activities of many AI companies. As you may recall from Chapter 1, Vladimir Putin has said that whoever leads AI will control the world. And, indeed, China plans to be that leader by 2030. The country plans an AI industry of $150 billion.

Among the many companies that are engaged in AI, three are investing billions of dollars, employing thousands of AI experts and robotic engineers, and acquiring global talents in AI. The three companies are Alibaba Group, Tencent, and Baidu. AI is already the priority of the Chinese government. In a cover story in Fortune, Lashinsky (2018) describes and analyzes the competition between Tencent and Alibaba.

TENCENT This giant e-commerce company has created a huge AI lab to manage its AI activities. The goal is to improve AI capabilities and support decision making in the following areas: computer vision, NLP, speech recognition, machine learning, and chatbots. AI is already embedded in over 100 Tencent products, including WeChat and QQ. A well-known AI slogan in China is “Juey, GO AI.” Tencent supports the robotic company UBTech Alpha. Tencent is the world’s largest Internet company, and AI im- proves its operations. Another slogan is “AI in all.” The company has a lab in Bellevue, Washington. Healthcare is a main research priority there. For more on AI at Tencent, see Marr (2018).

762 Part V • Caveats of Analytics and AI

Alibaba has developed a cloud-based model known as ET Brain alibabacloud.com/et. The logic is that today and in the near future, we are and will be doing business in the cloud computing environment. Content, knowledge, and data are in the cloud, and Alibaba is both a user and a pro- vider of iCloud. The ET Brain model is illustrated in Figure 14.6.

ETBrain consists of three parts: technolo- gies, capabilities, and applications. Technologies include Big Data and analytic processing, neu- ral networks, video recognition analysis, and

machine learning. These technologies provide four major capabilities such as cognitive percep- tion, reasoning, real-time decision making, and machine learning (see the middle level in the figure). The capabilities drive a large amount of applications, such as e-commerce activities (both business-to-business and business-to-consumers), medical and health care, smart cities, agricul- ture, travel, finance, and aviation. All-in-all, it is a super-intelligent AI platform. The ET Brain is illustrated in a 26:29 min . video at youtube.com/ watch?v=QmkPDtQTarY.

Application Case 14.3 How Alibaba.com Is Conducting AI

BAIDU Baidu started NLP research five years before Google to improve its search engine capabilities. The company is located in the Silicon Valley, Seattle, and Beijing. Baidu has sev- eral products. One is Duer OS, a voice assistant that is embedded in more than 100 brands of appliances in several countries. The product is now optimized for smartphones. Baidu is also working on autonomous vehicles. Finally, the company promotes facial recognition in the en- terprise (replacing ID badges). Baidu’s AI is growing but still much smaller than that of Alibaba.

ALIBABA The world’s largest e-commerce company and the provider of cloud comput- ing and IoT platforms, Alibaba is active in AI projects and is an investor in AI companies, such as in the face recognition giant SenseTime. Alibaba has developed a methodology for conducting AI, which is described in Application Case 14.3.

Int ell

ige nt

AI

fou nda

tio n Int

elli gen

t

thi ngs

Int elli

gen t

app s,

ana lyti

cs

Digitaltwins Digital

Conversational platforms

Immerse experience

Cloud to theedge

Mash

Blockchain

Event driven

Adaptive risk and trust

FIGURE 14.5 Gartner Prediction. Drawn by E. Turban

Chapter 14 • Implementation Issues: From Ethics and Privacy to Organizational and Societal Impacts 763

Technologies

Big data processing Multi-faced

security

Neural networks Real-time analytics

Video recognition and analysis

Advance data processing

Brain family : city brain : industrial brain : environmental brain aviation brain : global data exchange brain : medical brain

ETBrain’s Capabilities

Reasoning

Real-time decision making

Machine Learning

Perceptual innovation

Cognitive Perception

Multi- dimensional awareness

Strategic decision making

Situational intelligence

Applications ; Innovations

Smart cities Travel Fashion Medical Image application

Environment Agriculture Retail Financial AI assistant

Aviation Transportation Voice recognition Facial recognition Text recognition

FIGURE 14.6 Alibaba’s ET Brain Model. Drawn by E. Turban. Based on text at Alibabacloud.com/et

Alibaba’s mission is to reach 2 billion con- sumers and to help 10 million businesses world- wide. To attain this mission, the company invested in seven research labs that focus on AI, machine learning, NLP, face (image) recogni- tion, and network security. Alibaba is using AI to optimize its supply chain, personalize recom- mendations, and provide virtual personal assis- tants. Alibaba concentrates on several industries and on AI-supported bricks-and-mortar shopping. For example, in its AI office in Hong Kong, the company opened “Fashion AI,” working with Guess Inc., helping shoppers to create an online ensemble while they are in a physical store. See engadget.com/2018/07/04/guess-alibaba-ai- fashion-store/. The company plans to rewire the

world with AI (see Knight, 2018) and may control the world commerce.

Questions for Case 14.3

1. Relate cloud computing to AI at Alibaba.

2. Explain the logic of the ET Brain model.

3. Search the Web to find recent Alibaba activities in the AI field.

4. Read Lashinsky (2018). Why is Alibaba in such strong competition with Tencent?

Sources: Compiled from W. Knight. (2018, March 7). “Inside the Chinese Lab That Plans to Rewire the World with AI.” MIT Technology Review; Marr, B. (2018, June 4). “Artificial Intelligence (AI) in China: The Amazing Ways Tencent Is Driving Its Adoption.” Forbes; A. Lashinsky. (2018, June 21). “Alibaba v. Tencent: The Battle for Supremacy in China.” Fortune. alibabacloud.com/et.

764 Part V • Caveats of Analytics and AI

The U.S.–China Competition: Who Will Control AI?

At the moment, U.S. companies are ahead of Chinese companies. However, this situation may be changed in the future due to the huge investments in AI in China and the support provided by the Chinese government. Note that a major topic in the U.S.–China trade ne- gotiations in 2018 centered on the use of technology by Chinese companies that employ U.S. knowledge and trade secrets.

The Largest Opportunity in Business

According to McCracken (2017), intelligent technologies provide the largest opportunity for tech companies since mobile computing. This is why tech giants and start-ups are trying to exploit AI. Desjardins (2017) provides an infographic about the future impact of AI that includes $15.7 trillion by 2030 in the form of productivity gains and increased consumer spending. By 2018, tech giants and others will invest $30 billion in research and development and $13.2 billion in start-ups. The largest improvement is expected in image and speech recognition products.

Note that despite their rivalry, Facebook, Amazon, Google, IBM, and Microsoft launched a partnership to research advancements and best practices in AI.

Conclusion

Now that you have completed reading this book you may ask, “What will happen to intelligent technologies in the future?” There will be a significant impact on business and quality of life. There will be changes, and they will be significant. With billions of dollars invested, mostly in AI, there will be advancements. Machines are getting smarter and smarter. For example, Alibaba’s copywriting machine, which is based on deep learning and NLP, can generate 20,000 lines of text in one second. The machine is so smart that it passed the Turing test (Chapter 2), which means that it is smart like a human but can work much faster. We will now look at two areas: business and quality of life.

IMPACT ON BUSINESS According to Kurzer (2017), there might be challenges, but AI was expected to flourish as of 2018. There is very little doubt that we will see increased commercialization of AI, especially in marketing, financial services, manufacturing, and IT support. For example, the quality and nature of the customer experience could be im- proved, augmented by AI applications, and IoT. Kurzer also predicted that there will be more proactive processes rather than reactive ones. There will be more people-machine collaboration and while many jobs will be automated, many new ones will be created. There is going to be more conversational AI due to the increased capabilities of chatbots and personal assistants such as Alexa, Siri, and Google Assistant. Gartner predicted that by the end of this decade, people will have more conversations with machines than with their immediate family members (gartner.com/smarterwithgartner/gartner- predicts-a- virtual-world-of-exponential-change/). Another area with promising ap- plications is image recognition. Google is a major force in both conversational and image recognition AI.

IMPACT ON QUALITY OF LIFE There will be impacts on life that will change the way we drive, eat, entertain, get services, learn, and fight.

A major area where AI intelligent systems have already made a stride is the health- care field. Bernard Tyson, CEO of Kaiser Permanente, made the following public state- ment: “I don’t think any physician should be practicing without AI assisting in their

Chapter 14 • Implementation Issues: From Ethics and Privacy to Organizational and Societal Impacts 765

Chapter Highlights

• Intelligent systems can affect organizations in many ways as stand-alone systems, or integrated among themselves or with other computer-based information systems.

• The impact of analytics on individuals varies—it can be positive, neutral, or negative.

• Serious legal issues may develop with the intro- duction of intelligent systems; liability and pri- vacy are the dominant problem areas.

• Many positive social implications can be expected from intelligent systems. These range from pro- viding opportunities to people to lead the fight against terrorism. Quality of life, both at work and at home, is likely to improve as a result of the use of these technologies. Of course, there are poten- tially negative issues to be concerned about.

• Growth of intelligent systems is going to lead to major changes in industry structure and future employment.

• A major battle is brewing about who owns the user data that are being generated from the use of smartphones, cars, and so on.

• In deploying intelligent systems, it is necessary to consider legal, privacy, and ethical issues.

• Placing robots as coworkers in the work force raises legal and ethical issues.

• Intelligent technologies may impact business pro- cesses, organizational structure, and management practices.

• It may be necessary to create independent orga- nizational units that deploy and manage intelli- gent systems.

• Intelligent systems may provide a considerable competitive advantage to their users.

• Intelligent systems may create massive unem- ployment mainly in routine and mid-management jobs.

• Eventually, intelligent system may cause unem- ployment even in skilled jobs. So retraining may be needed.

• Intelligent systems may result in restructur- ing many jobs notably through human-machine collaboration.

• Intelligent systems will create many new jobs that require specialized training.

• The use of intelligent systems automation may result in a shorter work week and a need to compensate those people who will lose their jobs.

• Some people are afraid of unintended conse- quences of having AI and robots. Machines will learn and may harm humans.

practice. It’s just impossible (otherwise) to pick up on patterns, to pick up on trends, to really monitor care.” Editors (2018) report that smart solutions can improve quality of life indicators by 10 to 30 percent. (The longer we wait, the higher the percentage will be.) Among the indicators that they cite are: having longer and healthier lives, reducing green- house gas emissions, saving 200,000 lives worldwide over 10 years (thanks to self-driving cars), reducing the commute time for people (fewer traffic problems), increasing the number of jobs (e.g., by new technologies and more productive business environments), and providing better and more affordable housing.

Autonomous vehicles, including drones, will clearly change our lives for the better, and robots will be able to serve us (especially people who are elderly and those that are sick), entertain us, and if properly managed, be our companions. For an impact of AI in the future on society, watch the video at youtube.com/watch?v=KZz6f-nCCN8/.

What will the unintended results be? What if robots will kill us all? Well, that prob- ably will never happen. People are smart enough to make sure that only good results will come from intelligent systems.

u SECTION 14.9 REVIEW QUESTIONS

1. Describe the AI activities of major U.S. tech companies. 2. Describe the work by Chinese giant companies. 3. Describe Alibaba’s approach to AI (The ET Brain model).