Read then write

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Carbots – Where are We?

Alice E. Fischer

August 30, 2017

Carbots. . . 1/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

The Vision and the Current Reality The Vision Current Reality

The Technology Hardware Technology Hardware Technology

The Issues Public Policy Where are we now?

Brave New World Cost Change

Carbots. . . 2/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Carbots – Where are We? Alice E. Fischer

The Reality

The TechnologyThe

Issues

The Dream

First version: June 1, 2016, Updated: July 28, 2017

Carbots. . . 3/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

The Vision

I. The Vision

Carbots. . . 4/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

The Vision

The Dream

1. Everyone will be safer because we will eliminate distracted driving, drunks, high-speed chases, etc. (Subtext: machines follow rules, humans don’t do so reliably.)

2. Children, the elderly, and partygoers can all get to their destinations safely without a competent driver. (Subtext: We want servants who can take care of the non-productive members of society.)

3. Our highways will be able to handle more cars, traveling closer together at high speeds, because a computer’s response time is faster than a human’s and can handle smaller distances safely.

Carbots. . . 5/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

The Vision

The Nightmares

The nightmares are the flip-side of the dreams:

1. It is clear that some people should NOT be driving cars, but will they be the ones using the carbots? Will we require that?

2. It is not at all clear that carbots will be safer; they are just different. A carbot may be safer in some ways, but also be more dangerous in other ways.

3. People will send children and the elderly to their destinations in carbots. Of course, we already have taxis and Uber for occasions when the competent adults are busy.

4. Any unexpected emergency might cause a crash. Would this involve a pile-up of a hundred carbots?

5. When a carbot starts behaving erratically, what next?

Carbots. . . 6/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Current Reality

II. Current Reality

Emergencies, Accidents, and Crashes Limitations

Carbots. . . 7/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Current Reality

Limitations as of August, 2014

1. The latest prototype had not been tested in heavy rain or snow due to safety concerns.

2. Because the cars rely primarily on pre-programmed route data, they do not obey temporary traffic lights or diversions.

3. In complex unmapped intersections. they revert to a slower “extra cautious” mode.

4. The vehicle has difficulty identifying when objects (trash and light debris) are harmless, causing the vehicle to veer unnecessarily.

5. The LIDAR technology cannot spot some potholes.

6. The LIDAR cannot discern when humans, such as a police officer, are signaling the car to stop.

Carbots. . . 8/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Current Reality

A Ticket

June 2015:

Google’s prototype carbot was being tested in California, with safety drivers aboard. Its speed was 25 mph in a 35 mph zone. As a consequence, one of the vehicles was stopped by police for impeding traffic flow. No ticket was issued.

Carbots. . . 9/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Current Reality

An Accident

June 2015:

A Google carbot and a carbot made by auto-parts maker Delphi had a near miss, but the cars managed to avoid each other at the last moment. The people in both cars had “minor whiplash.” The Google car was just scraped; the Delphi car lost its bumper. There was no police report.

Carbots. . . 10/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Current Reality

Crashes

In July 2015, Google confirmed that its 23 self-driving cars had been involved in 14 minor collisions on public roads.

I Nine involved being rear-ended at a stop sign or traffic light. (Sudden, unexpected stops?)

I In two, the vehicle was side-swiped by another driver,

I One involved another driver rolling through a stop sign,

I In one, a Google employee was manually driving the car.

Google maintains that the carbots were not at fault.

Carbots. . . 11/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Current Reality

Testing, Jan 2016

I Google’s vehicles have driven driven nearly 1.5 million miles autonomously. They have gone up San Francisco’s Lombard Street (steep hairpin turns), through city traffic, over the Golden Gate Bridge, and around Lake Tahoe.

I New data from Google show that autonomous cars are making progress. There was an overall decline in incidents in which the technology failed since Fall, 2014.

I The 53 carbots have been in 17 crashes, but were never considered “at fault” until recently. Most of the accidents were rear-end collisions.

I Experts cautioned that the company’s reports do not provide enough information to definitively say whether the technology is safe.

Carbots. . . 12/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Current Reality

Some Emergencies

Jan 2016:

Google revealed that during 272 occasions over 14 months, drivers took control of autonomous vehicles because the software was failing.

In 69 other incidents, the test drivers chose to take control of the autonomous vehicles to ensure that the vehicles operated safely.

Carbots. . . 13/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Current Reality

The first crash clearly caused by the carbot, February, 2016

I 3:20 p.m. Feb. 14, about three miles from Google’s headquarters in Mountain View, Calif.

I Google’s car was attempting to make a right-hand turn on red and moved to the right side of a wide lane to pass traffic stopped at the light. But as the carbot neared the corner, its path was blocked by sandbags around a storm drain.

I The carbot tried to go around the sandbags by cutting into the lane of vehicles on the left. It then struck a metal piece connecting the two parts of an accordion-style bus on the other road.

I The carbot was going less than 2 mph and the bus was moving at 15 mph. There were no injuries in either vehicle and the crash was described as minor.

Carbots. . . 14/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Current Reality

Google’s Response !!

I Google’s driver thought the bus would slow or stop so the Google car could go ahead of it. Google characterized the crash as a misunderstanding and a learning experience, saying its cars will learn that large vehicles are less likely to yield than other types of vehicles.

I “We hope to handle situations like this more gracefully in the future,” Google said. “We clearly bear some responsibility [!!], because if our car hadn’t moved [made an inappropriate right turn on the red] there wouldn’t have been a collision.

I Google said “This type of misunderstanding happens between human drivers on the road every day”.

Carbots. . . 15/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Current Reality

The Tesla Crash.

I The carbot marketed by Tesla relies only on visual systems for road-sense. (No radar, no lidar.)

I In May, 2016, a Tesla car on autopilot crashed and killed the driver.

I The car hit an “invisible” truck that was crossing the road. It went under the truck and kept going some distance. The owner was decapitiated.

Carbots. . . 16/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Current Reality

Tesla’s Response. I Tesla markets the self-driving feature as a safety assist, not as

a substitute for a driver. The “driver” was supposed to be driving, with his hands on the wheel.

I However, during the 37 minutes of the trip, his hands were on the wheel for only 25 seconds.

I The car’s warning system gave him flasing warnings seven times, which were ignored.

I So it is not Tesla’s fault, and there is nothing wrong with the car except that it cannot recognize a truck driving across the road. (!!!)

I A friend of a friend at Google says that Tesla is giving carbots a bad name and Tesla is irresponsible for using only a vision-based system.

Carbots. . . 17/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

III. The Technology

Hardware Technology Software Technology

Carbots. . . 18/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Hardware Technology

Hardware Technology.

The LIDAR on top allows the vehicle to generate a detailed 3D map of its environment. The car combines the 3-D maps from the laser system with high-resolution maps of the world, producing different types of data models that allow it to drive itself.

The system works with a very high definition inch-precision map of the area the vehicle is expected to use, including how high the traffic lights are; in addition to on-board systems, some computation is performed on remote computer farms.

Is this scalable? Will a road full of carbots interfere with each other?

Carbots. . . 19/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Hardware Technology

Hardware Technology.

Different types of cars have been retrofitted with self-driving equipment, including the Toyota Prius, Audi TT, and Lexus RX450h.

In May 2014, a prototype of a 100% driverless car was introduced, with no steering wheel or pedals.

Carbots. . . 20/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Hardware Technology

No steering wheel, no pedals. No panic button?

1. The 64-beam laser range finder on top scans 360 degrees.

2. Radar on the front measures the speed of the vehicles ahead.

3. An orientation sensor tracks the car’s motion and balance.

4. A wheel-hub sensor detects the number of rotations to help determine the car’s location.

5. A processor reads the data and regulates vehicle behavior.

Carbots. . . 21/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Hardware Technology

Google’s 2016 Patent Human flypaper, to grab and hold a victim after a crash.

Carbots. . . 22/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Hardware Technology

Software Technology.

Because the software is a trade secret, we cannot know what is actually being used. However, these technologies are almost certainly used:

I Expert Systems.

I Deep Learning.

I Mapping Algorithms.

I Modeling.

Carbots. . . 23/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Hardware Technology

Expert Systems.

An expert system solves complex problems by reasoning about knowledge. It has an inference engine and a knowledge base of facts and rules. The inference engine applies the rules to the facts to deduce new facts and make decisions.

Expert systems were among the first truly successful forms of AI software and are widely used today for a variety of applications:

I Chatbots: the ES tries to discern the meaning of the human input and produce a relevant response.

I Medical diagnosis: the ES correlates symptoms with a huge database of possible conditions to propose a diagnosis.

I Process control: the ES interprets, predicts, corrects, and monitors system behaviors.

Carbots. . . 24/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Hardware Technology

Mapping Algorithms.

3-D mapping hardware/software is a central part of a Google carbot system. Massive amounts of data coming from the LIDAR hardware must be analyzed in real time, mapped onto a model, and then interpreted.

The interpretation determines whether the carbot continues, turns, swerves, or stops. The algorithms for finding edges and objects are old and well studied.

Today’s vision systems almost certainly involve both algorithmic parts and neural nets (deep learning).

Carbots. . . 25/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Hardware Technology

Deep Learning. Deep learning (formerly called a “neural networks” attempts to model high-level abstractions by using multiple processing layers, with complex structures composed of multiple non-linear transformations.

The goal is to replace handcrafted representation of models with efficient non-algorithms for unsupervised or semi-supervised feature learning and extraction. Success relies on use of extensive training cases that cover all the situations needed.

There are various deep learning architectures based on deep neural networks. These have been applied with great success to fields like computer vision, automatic speech recognition, natural language processing, audio recognition and bioinformatics.

Carbots. . . 26/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Hardware Technology

Modeling. A model is an abstract representation of a piece of reality. It captures some important features and necessarily ignores others.

In any software system, the quality of the models is critical to the quality of the outcomes.

All AI software exists in relation to a model of reality. The models guide the training of the AI.

If a concept, situation, or object is missing from the model, it will not be identified and interpreted by the AI system.

Example: today’s carbots do not obey a policeman with a stop sign at a construction site because the site and the human are not on the map (in the model). The carbot will “see” something, but be unable to interpret it properly.

Carbots. . . 27/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Hardware Technology

A Carbot creates a model. The models needed to control a car are extremely complex. and are formed by combining a detailed map of the road with the output from the lidar system.

Extensive work has been done and driverless cars can usually navigate around California without having or causing problems.

In the graphic, this completed work is represented by the yellow and lavender bands, all built on the red core of preexisting technologies.

Core hard and soft technologies

Basic driving: Steering, speed regulation, handling intersections, following a route. Stop light placement, bicycles, avoiding collisions, traffic, freeways, night driving. Rain, fog, snow, ice, mountains, potholes, portable stop signs, sandbags, predicting other drivers, etc, etc. etc.

Carbots. . . 28/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Hardware Technology

The Unfinished Modeling / Training / Testing

Google acknowledges that it currently has a varied list of unsolved problems. In the graphic, they are colored blue-green.

These problems are less central to the driving and less usual, so it has been possible for Google to field a small fleet of driverless cars without experiencing a lot of crashes.

The problem is, there are a huge number of unsolved issues (known unknowns and unknown unknowns). In fact there is an unending stream of unsolved problems.

Example: we do not know whether a carbot that functions well in the daytime in California on a dry day will function safely in Connecticut in a heavy rainstorm at night.

Carbots. . . 29/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Hardware Technology

AI’s are Unpredictable The interaction of this large number of unsolved problems with deep learning methods is troublesome.

I Each added rule or training case has the potential to cause novel behavior in previously tested situations.

I As more and more training cases, facts, and rules are added to the AI, the likelihood of unexpected bad decisions and other software malfunctions (freezing up) increases.

I Because of the nature of the AI, nobody can know what caused a bad outcome, and nobody can identify a way to fix it other than adding more rules and training cases.

I The process of retraining the AI is likely to go on for many years, with the carbots getting better and better, but never quite behaving sensibly in untested situations.

Carbots. . . 30/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

IV. The Issues

Cost Public Policy

Where are we now? Brave New World

Carbots. . . 31/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Public Policy

Urgent: Start Now

I The laws governing human drivers are not adequate.

I We HAVE NO nationally accepted public policy regarding standards for or regulations governing carbots.

I Such policy needs to be in place before we begin to convert our present way of life to a very different one.

I The problem is that nobody in power has yet begun the necessary national exploration of these questions: not government, not industry, not techies.

I The legislators who are going to make fundamental decisions in this case are not well informed about the realities.

I The scientists and ethicists who should inform the legislators have not been identified.

Carbots. . . 32/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Public Policy

Missing: the Three Laws of Carbotics

We need to formulate the “three laws of carbotics”. We need an accepted, general, national idea of the limits within which these vehicles will operate and the goals they should strive to achieve.

It is not enough to have a patchwork of laws, failing to cover the whole picture and differing from one state to the next.

For example, in an emergency situation, who should the carbot seek to protect first?

I The unprotected pedestrian with a stroller?

I The occupants of the carbot?

I The physical integrity of the carbot?

Carbots. . . 33/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Public Policy

Oversell.

I “Only one accident in 1,500,000 miles of autonomous driving” was caused by the carbot. The question is, how many of the other accidents that DID happen would have been avoided by a human being?

I . . . And what about the 341 times that the “safety” driver intervened to avoid a crash

I . . . And what about the incidents that were unreported because they involved only minor property-damage?

I We simply lack the information to know how good a driver the Google carbot actually is.

Carbots. . . 34/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Public Policy

Oversell.

I “Safer than humans”? Some humans yes, and others, no.

I This promotional slogan has had the bad effect of making many educated people believe that the problems are all solved.

I My husband and I have been driving for 2 times 59 years at about 20,000 miles per year (2.36 million miles). Our record: 11 collisions. Two were trivial were not reported. Four happened while we were stopped at a light. Once we hit a possum on the road at night. Once we were sideswiped by someone passing on the right. Once we were hit by a car coming out of a parking lot. If we were Google, only two of these incidents would have been reported to the police.

I If you compare our record to the record of the Google car, Google is not impressive.

Carbots. . . 35/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Where are we now?

The Invisible Truck. This happened on Rte. 15 in Meriden, a road that is very familiar to me. I was a passenger in the front seat.

I I noticed that there was something strange about the road ahead. I saw road, trees, and sky, but the picture did not match my memory of that place.

I I paid closer attention, trying to figure out what I was seeing. Soon, the image made sense. The “sky” was the side of a tall white truck, turning onto the road, and momentarily broadside to the direction of traffic.

I We slowed down, the truck continued its turn, and everything was normal.

This is clearly what happened to the Tesla car before the fatal crash. The difference? The Tesla did not recognize that something was abnormal.

Carbots. . . 36/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Where are we now?

A human can often avoid an accident. These situations did not result in an accident, primarily because of human perception and comprehension.

I (I-95) My view ahead was limited by traffic. I saw person walking along the shoulder in the wrong direction and knew something unusual was happening. Seconds later, the traffic cleared and I saw an extension ladder across one lane and extending into my lane. I was able to veer and only clipped the corner of the ladder.

I (I-91) Approaching a construction zone, I saw a car going too fast toward a construction zone. I watched as it hit a parked truck and the truck began to roll across all four lanes, on a collision course with me. I accelerated a bit and it passed behind me.

Carbots. . . 37/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Where are we now?

Case study.

While driving on Ridge Road last week we saw a pedestrian approaching us at the edge of our travel lane, pushing a shopping cart.

I Mike pulled gently across the center line to give him a wide berth. (There was no oncoming traffic.)

I But what if there was, and there was no space for two vehicles and a cart to pass? What should happen?

I Mike says that he would stop. This protects all of the people, but risks a rear-end collision. Should a carbot be expected to stop instead of veering?

Carbots. . . 38/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Where are we now?

People Understand Other People; Carbots Don’t

Sometimes a warning is small, like a bouncing ball in the road:

I (State Street) Mike glimpsed a car on a side street briefly before the view was occluded by a stopped bus.

I Out of sight, the car reached the corner and turned right, around the bus, which filled much of the southbound travel lane.

I The turning car went well over the yellow line and headed directly toward Mike.

I Mike was already slowing down and squeezing right because of the brief warning.

I There was no head-on collision.

Carbots. . . 39/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Where are we now?

Trust.

Google projects having the technical issues fixed by 2020 and is lobbying now to allow the sale of carbots.

I Google is saying “trust us. Is that a reasonable request?

I How well can you trust anything that is not open to public scrutiny?1

I Google has not been fully open and honest about the track record of their cars.

1This question about carbots has an analog in the ongoing battle over voting machines and online voting.

Carbots. . . 40/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Where are we now?

Security. How safe are these systems from takeover by hackers? Of course, we don’t know.

I The public deserves to be informed about the security measures that have been implemented.

I A segment on “60 Minutes” (2015) featured a reporter losing control over her cars horn and brakes as a hacker clicked away on a nearby laptop.

I WIRED magazine staged a takeover demonstration in July 2015. The zero-day exploit let hackers send commands through the internet to the Jeep;s entertainment system, then to its dashboard functions, steering, brakes, and transmission. The hackers were miles away and the driver was completely helpless.

Carbots. . . 41/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Where are we now?

Liability Law.

When the first death happens, who will be legally liable?

I We HAVE NO definition of who will be responsible when a carbot crashes and injures someone.

I The ordinary driver liability laws do not apply.

I At least one state has a law that exempts the original manufacturer of the car from liability after it has been retrofitted to be driverless.

I The nature of possible sanctions is unclear: You can’t just revoke the license of a carbot after an accident – there are presumably hundreds of others just like it.

I The nature of possible remedies is unclear.

Carbots. . . 42/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Where are we now?

Legislation

Public road testing of driverless cars is legal in California, Michigan, Florida, Nevada, Arizona, North Dakota, Tennessee, the District of Columbia, and Utah. (Updated 23 Mar 2016)

I In California, all cars must have licensed drivers, whether or not they are driving.

I In Michigan, GM built an entire small town as a test set for carbots.

I Lobbying by Google in the US Senate is underway to change the national law. Google is pushing Congress to give the NHTSA new powers to grant it special, expedited permission to sell cars without steering wheels or pedals.

Carbots. . . 43/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Where are we now?

Lobbying

Currently, states are being pressured one at a time to allow testing, and they are passing idiosyncratic laws.

“Google 2 expected the California Department of Motor Vehicles to release precedent-setting regulations regarding driverless cars in January 2015, allowing the company to give public access to the prototypes. However, as of November 2015, the DMV still has not done so.

Google - and other companies - are not ready to share trade-secret safety data, although they continue to push for the publication of California regulations.”

2California’s Red Tape Slows Google’s Self-Driving Roll, www.yahoo.com, Retrieved 2015-11-16.

Carbots. . . 44/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Where are we now?

California is not Sold. In December 2015, the California Department of Motor Vehicles issued long-awaited proposed regulations governing autonomous vehicles and invited public comments on the draft regulations.

If adopted, they would require self-driving cars to have a steering wheel and pedals, and a human driver onboard who holds an “autonomous vehicle operator certificate”. They would also hold the occupant responsible for accidents and violations of traffic laws, whether or not the driver was at the wheel.

The DMV stated, “Given the potential risks associated with deployment of such a new technology, [we believe] that manufacturers need to obtain more experience in testing driverless vehicles on public roads prior to making this technology available to the general public.”

Carbots. . . 45/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Cost

Equipment Cost.

Last year, Googles cars cost about $320,000, including the basic Prius, a $70,000 LIDAR system, a $10,000 radar system, and a $200,000 GPS array. This does not include the cost of the underlying car.

Decreases in the cost of the technology are predicted. For example, Velodyne has introduced a new, smaller LIDAR system that costs only $5000.

One estimate is that the equipment will add $7000 to $10,000 to a cars sticker price in 2025. This puts a carbot out of the reach of most citizens.

Carbots. . . 46/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Cost

Social Cost

Large-volume sales will be necessary to bring the price down to anything an ordinary worker can afford.

There is some hope that gradual market penetration and large-scale production will reduce that cost to $3000 or $5000 in 2030 to 2035. This will be a problem for many people who now own cars: these cars will be substantially more expensive to buy and to maintain.

The added safety that people expect carbots to bring, will only happen when no traditional cars are on the road with human drivers. Until then, the carbots and the humans will endanger each other in new and untested ways.

Carbots. . . 47/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Change

Cars vs. Carbots

Assume that someday it becomes legal to sell carbots.

I Some people will resist giving up their cars. The greatest resistance will come from the poor.

I Aggressive drivers will also resist the change. (I know a person who routinely drives 80 mph on freeways.)

I There are people who will never trust robots. Maybe rightly.

I Dangerous drivers will still be on the road unless the courts revoke their licenses.

I At least a carbot won’t direct road-rage against others!

I The question is, will they be a magnet for the road-rage of the most dangerous class of drivers?

Carbots. . . 48/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Change

The Weather “If you don’t like the weather in New England now, just wait a few minutes.” – Mark Twain

I I ride daily in a car that has “eyesight” connected to the brakes. It will slow down if you get too close to a car ahead.

I When it was new, 18 months ago, it would suddenly stop working in the middle of the trip. Two firmware updates seem to have fixed this. (There were no recalls, though.)

I Now it works well as long as there is no fog or heavy rain. It can’t “see” under those conditions.

I I would not want to rely on a machine that fails as soon as the going gets rough.

I Suppose a carbot checks weather conditions on the internet, sees that they are bad, and decides not to go. What then? What if it is an emergency and the risk is warranted?

Carbots. . . 49/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Change

This is a disruptive technology. It will change many aspects of life for ordinary people.

I Getting a driver’s license is a rite of passage for our youth. It is the first time they have had total responsibility for their own safety and that of others. It is the first way in which a teen becomes like an adult.

I These high-tech cars are expensive – extremely so. Car ownership could become just one more thing that the poor and the middle class cannot afford. Would they then call taxis? This would continue to exacerbate the imbalance of power between the 1% and the 99%.

I How will radar and lasers in continuous use affect health? I The advantages of carbots will not be realized as long as they

share the roads with cars. At some point, the right of an individual to choose to own an ordinary car may be at risk.

Carbots. . . 50/51

Outline The Vision and the Current Reality The Technology The Issues Brave New World

Change

Resources Many thanks to Wikipedia and several news media for the articles from which this information was gleaned:

I https://en.wikipedia.org/wiki/Deep_learning I https://en.wikipedia.org/wiki/Expert_system I https://en.wikipedia.org/wiki/Google_self-driving_car I http://money.cnn.com/2015/07/17/autos/google-self-

driving-car-injury-accident/index.html I http://arstechnica.com/cars/2016/02/google-self-driving-car-

strikes-bus-in-california/ I California’s Red Tape Slows Google’s Self-Driving Roll,

www.yahoo.com. Retrieved 2015-11-16. I https://www.yahoo.com/tech/can-hackers-really-take-over-

your-car-110827803044.html

My apologies if I have omitted some of the sources. Carbots. . . 51/51