In a popular children’s book called If I Built a Car, a fanciful fledgling engineer (who is probably about ten) waxes enthusiastically about designing a car that houses an onboard swimming pool, makes milk shakes, and can both fly and dive under water.1 Of course, the car has a robot driver that can take over if the humans need a snooze.
We aren’t getting cars that can make milk shakes or are big enough to house a decent sized swimming pool, and flying cars remain a couple of decades away. But our robot drivers are here.
There are debates in mainstream media over whether driverless cars will ever be adopted and whether we can trust our lives to a machine. A survey by the American Automobile Association in March 2016 revealed that three out of four U.S. drivers would feel “afraid” to ride in self-driving cars, and that just one in five would entrust his or her life to a driverless vehicle.2
When I first encountered the Google car in Mountain View, back in 2014, I had the same doubts. If I had taken the survey, I would have been in the three out of four who are afraid. And then, in July 2016, I took delivery of a new Tesla that had some of these self-driving capabilities.
At first, the thought of letting my car drive itself was indeed frightening. But the highway was almost empty, and the lanes were clearly marked, so I took the risk and engaged the autopilot function. I kept a firm grip on the wheel, because I didn’t want to put my life in the hands of software. The fear lasted for just five minutes. Curiosity got the better of me, and I let go of the steering wheel to see what would happen. The car continued to drive just fine; it didn’t need me. Twenty minutes later, I had one hand on the wheel and I was checking e-mail with the other as the car did the driving for me. I did take full control when the road was narrow or the terrain was uneven, but, by and large, I became as comfortable with the car’s autosteer function as I am with cruise control.
Tesla’s autosteer is just one step on the path to a fully autonomous car, but, as other technologies subject to Moore’s Law are doing, self-driving systems are becoming exponentially more functional. All of the Teslas on the road are learning in tandem; they have collectively traveled in the millions of miles. Within three to four years, my Tesla will be driving by itself without any help from me.
This is not to say accidents won’t happen. One already did, and it was a big deal; a man in Florida using Tesla’s autosteer crashed into a turning truck.3 He was killed. But compare this to the many thousands of fatal accidents that happen due to drunk driving and, at least statistically, we would all be safer if we used autosteer, even in its current less-than-perfect state.
What seems hard to do today will seem trivial in a few years, as we replace big, clunky, expensive systems with tiny, reliable, affordable autonomous software and vehicle-sensor packages. I expect that you will get used to this about as fast as I did. It will be an amazing transition, and we won’t want to look back.
Few people seem to fully grasp the profound improvements in our lives that driverless cars will bring. Their adoption will slash accident and fatality rates, saving millions of lives. As well, it will remove one-third to one-half of all vehicles from city streets. A large percentage of the cars on the streets of New York, San Francisco, and London at any one time are looking for parking; but self-driving cars don’t need to park: they can continuously circulate, picking up and dropping off passengers. The Earth Institute at Columbia University projects a 75 percent reduction in the cost of car ownership, because fewer shared vehicles will be necessary to provide the same service collectively that personally owned vehicles provide.4 During peak hours, those shared vehicles will be in use 90 percent of the time. And, with no more need for steering wheels and other systems enabling human control, vehicles will be lighter and far more fuel efficient. Most important, car sharing will cost a fraction of what car ownership today costs. Owning a car for daily, personal transportation will seem impractical.
Self-driving cars will also deliver incontrovertible social benefits. With self-driving cars, the disabled will no longer struggle to find transportation; they will have an on-demand personal driver. Several years ago, as the New York Times in November 2014 relates it, Google’s self-driving car team contacted Steve Mahan, Executive Director of the Santa Clara Valley Blind Center.5 The team wanted feedback and let Mahan come along for test drives in earlier self-driving Prius models as well as in the latest Google car. “My experience with Google has been terrific, and I want it to happen,” Mahan told the Times. “Everyone in the blind community wants it to happen.”
Other groups will also benefit in tangible ways. Women and children will never worry about getting a cab ride late at night. Once all drivers are off the road, traffic violations will no longer be an issue, and cops will have fewer reasons to pull over cars, which should reduce instances of the currently vicious discrimination against individuals “driving while black.” Teens will not face insurance discrimination as they do today, and their parents will not have to pay for the dubious privilege of teaching a teenager to drive. People living in the country will finally gain access to transportation services that put them nearly on par with their city cousins. Pedestrians will stop worrying about getting hit by cars in intersections.
Let me paint a picture of what streets will look like in an age of driverless cars. We will no longer need traffic lights: robot cars will synchronize wirelessly to time mass movements across city intersections and entries onto freeways or balletic dances around four-way stop signs. Having no human eyes behind the wheel will obviate much of the need for signaling and signage. When all the driverless cars are talking to each other, there will be no need for them to ever come to a complete halt and waste all their kinetic energy.
So we will be able to forget traffic lights—and stop signs, yield signs, lighting on freeways, and dozens of other transportation-infrastructure elements catering to human drivers. This great elimination will save many, many billions of dollars in the United States. Equally important, self-driving cars will eliminate the need to build these types of infrastructure in less developed countries in which traffic lights, freeways, and other modern traffic-control features have yet to be put in place. The future cost savings to those countries will be astronomical. In that future, the benefits of self-driving cars will be far more evenly distributed.
Eliminating human drivers will also allow automobile designers to build cars from a completely different mindset. Driverless cars will not need steering columns, brake pedals, accelerator pedals, or any of the other components drivers use for slowing or accelerating. They will not need a gearshift panel in the middle of the driver compartment or an emergency brake pedal. The A.I. system driving the car will also reduce accidents to negligible levels. Once accidents cease, there will be no heavy steel protective beams in the doors, or crumple zones. Self-driving cars will not require bumpers, seat-belt assemblages, or bulky airbags.
Dropping all of this extra mass and complexity will allow cars to be two things that we love: super efficient and super fast. Both will be important in the driverless future. Today a Tesla can drive 300-odd miles on a single charge. But eliminating its human-centric components would likely extend its charge range significantly.
Forgoing all those extra components could also make room for other things. Want to work while driving? You can put up a wide-screen thin plastic organic L.E.D. display and tap away at a full lap desk. Need a nap? Your car seat will recline all the way if there is no one sitting behind you, allowing full repose.
Another option would be to power a lot more enclosed space with the same drive-train and motor. Imagine if you could double the size of a motorhome: why own a real house, if living on the road is no longer inconvenient, uncomfortable, or cramped?
And let’s consider speed. Since humans will not be driving, and A.I. will keep all the cars moving in perfect order, cars could move at ludicrous speeds, to steal a phrase from the Tesla folks.
That could greatly affect the way we live and the way we think about cities. Right now, there is a schism in the Bay Area, between San Francisco and the rest of Silicon Valley. Many companies are moving up to San Francisco because that’s where their workforce wants to live. Driving from San Francisco to Palo Alto can often take one and a half hours due to nasty traffic. Commuter train service is jam packed and also subject to delays. As a result, I see my friends in the city less and less. And the mixing and melding of ideas that comes from people’s ability to move and meet suffers.
If a self-driving Tesla could move at up to 200 miles per hour on the highway and drive without stopping, a door-to-door commute from San Francisco to Palo Alto would likely be less than fifteen minutes (it’s only a thirty-odd-mile distance), and the problem would be solved. Middle-class workers in New York City, priced out of the inner city, could live at the beach, in Far Rockaway, and make it to Manhattan in around ten minutes.
The economic benefits, of course, will be massive as well. We will spend less time trapped behind the wheel and more time doing creative things. Cities can free up parking spaces and parking garages for apartments. It will be like it was to get turn-by-turn GPS: once we had it, we couldn’t understand how we had lived without it.
My grandchildren will ask me to tell them what it was like to drive a car in an old city. I’ll tell them it was scary, dangerous, and wasteful, and that they are lucky to have a better way of living. Oh, and, by the way, those Google cars we’ve been talking about have yet, despite millions of miles traveled on the roads, to cause a single fatal or serious accident. The few minor accidents that did happen were because of the pesky, ill-mannered, and dangerous humans that they had to share the road with.
Even after decades of decline, car accidents are a leading cause of preventable death in the United States. The National Highway Transportation Safety Agency surveyed tens of thousands of crashes and found that human error was a probable cause in 92.6 percent of them.6 Worldwide, 1.25 million people died in car accidents in 2013, according to the World Health Organization.7 In the United States in 2013, more than 32,000 people died in car crashes, despite access to one of the best emergency-healthcare systems in the world.8 In the developing world and middle-income nations, the road-traffic death rates are twice as high as in most developed Western countries, partly because of the difficulty of getting the casualties to hospitals in the critical first hour after the accident from locations where transportation infrastructure is lacking.9 If all of those cars outside the United States had been self-driving, in all likelihood we would have avoided 95 percent or more of those accidents and saved more than a million lives each year.
Simply put, people are poorly designed to guide two-ton hunks of metal. People drink and drive, fiddle with radios, fall asleep at the wheel, drive too fast, mistake the accelerator for the brake, and on and on and on. Many crashes are blamed on inattention. In other words, people cannot pay attention well enough to not crash. Autonomous vehicles are subject to none of these problems. The cost of self-driving systems is rapidly falling, and within a decade will be no more than $100. So there is a clear imperative to adopt self-driving cars as quickly as possible.
In the United States, big-rig trucks remain a major source of fatalities, demolishing cars in accidents. Many of these accidents happen on interstate highways, where drivers sit behind the wheel for days and frequently flout regulations on driver hours. A significant portion of these accidents, most of which are fatal to car drivers, are caused by the truck drivers falling asleep or driving with serious sleep deprivation. So it was no surprise when Daimler Benz put the first self-driving big-rig on the road in May 2015. Approved for use in the state of Nevada, the truck will handle driving duties on highways, ceding city driving to a human driver, who remains on board at all times.
Driverless cars also provide safe options for getting home at night. Instead of having to pay for a taxi or Uber when getting home late, women (especially young women) can summon a self-driving car.
In this discussion, I’ve focused on the United States. But the benefits are even greater to the crowded and polluted cities of the developing world, which will benefit tremendously from dramatically lower energy usage, an inexpensive transportation system for all, and a reduction in traffic and smog.
The United States has no monopoly on this innovation, and China may well leap ahead. Its leading technology company, Baidu, has developed its own self-driving software. After testing its technology in Beijing and Wuhu, in China’s southeastern Anhui province, Baidu obtained permission, in September 2016, from the state of California to begin testing there.10 It won’t surprise me if Baidu perfects its software before Google and Tesla do and China starts transforming entire cities into “autonomous car– only” zones.
And in August 2016, the world’s first self-driving taxis were picking up passengers in Singapore. This was in a 2.5-square-mile business and residential district called one-north. Singapore being a small, landlocked island with congested roads, its transport planners are very motivated to replace its cars with robots. “We face constraints in land and manpower. We want to take advantage of self-driving technology to overcome such constraints, and in particular to introduce new mobility concepts which could bring about transformational improvements to public transport in Singapore,” said Pang Kin Keong, Singapore’s Permanent Secretary for Transport, to the Associated Press.11
Mandating autopilot for everything that moves would devastate employment in the sector. According to the American Trucking Associations, in 2010 approximately 3 million truck drivers were employed in the United States, and 6.8 million others were employed in jobs relating to trucking activity, including manufacturing trucks, servicing trucks, and other types of jobs.12 So roughly one of every fifteen workers in the country is employed in the trucking business.
According to the U.S. Bureau of Labor Statistics, roughly another 300,000 people work as taxi drivers and chauffeurs. Those numbers would probably swell considerably if they included the new wave of part-time drivers. Uber, for example, claims over 14,000 cars in New York City alone.
For the near future, job growth in these industries is quite strong. But over time, driverless vehicles would mean the loss of close to 5 million jobs to the robots, with no obvious replacement jobs in sight.
And even though it is a near certainty that replacing human-directed vehicles with self-driving vehicles would reduce casualties, we already know that humans are more likely to blame robots when things go wrong but less likely to credit them for improvements. Researchers Tammie Kim from M.I.T. and Pamela Hinds from Stanford, examining this topic specifically, wrote in a paper titled Who Should I Blame? Effects of Autonomy and Transparency on Attributions in Human-Robot Interaction:
Our results suggest that when a robot has more autonomy, people will attribute more blame to the robot and less to themselves and their co-workers. This is consistent with our prediction that autonomy will contribute to a shift in responsibility from the person to the robot. It is interesting to note, however, that attributions of credit did not show the same pattern. That is, people shifted blame for errors, but not credit for successes to the robot.13
And then there are the impacts on our cities, social structure, and industries.
When parking decks become freed up and streets become pedestrian walkways, city layout becomes more flexible. We will be able to set large parts of the cities aside for parks and recreation. With location and distance not being barriers, we will be able to live anywhere, and patterns of social interaction will change. Imagine being able to visit friends for dinner who live in a nearby city or being able to get to the beach on a weekend without any traffic delays.
The real-estate industry will surely be in turmoil as land-use patterns change and reverse urbanization happens. The industry won’t be able to predict space utilization, because there are no precedents for the type of changes that will be occurring in the mid-2020s.
The automotive industry will be in decline because the number of cars purchased—even by the car-sharing companies—will fall dramatically. And then what happens to the car dealerships? When we can get from one city to another in relatively short periods in comfort in an autonomous car, why would we bother to take the train or struggle with the long security check lines at airports? For me, it’s already a toss-up between driving and flying when I want to travel from San Francisco to Santa Barbara, which is four and a half hours away by car and takes four hours by plane and taxis (provided there are no flight delays). The self-driving cars will easily tip the balance; for any trips on the West coast, I’ll forgo the flights. Imagine the disruptions to the railroad and airline industries when we all start making this choice.
And all of this begins to happen by the early 2020s. If I can rely on Elon Musk, my Tesla will become fully autonomous as early as 2018; 14 and Uber’s CEO, Travis Kalanick, has signed a pact with Volvo to have self-driving cars on the roads by 2021.15
I simply can’t wait for self-driving cars to take over our roads; I see them as increasing our personal autonomy as much as, if not more than, anything else discussed in this book. Let’s be honest: we may think we own our cars, but in reality our cars also own us. Buying a car is one of the most stressful processes in our lives. Fixing a car (or finding mechanics we can trust) is an equally problematic and far more common problem. Managing our auto insurance, washing our cars, maintaining our cars, and then ultimately getting rid of (selling or donating) our cars all take big chunks out of our lives. Then there is the part of our lives lost to driving a car in unpleasant circumstances, such as fighting rush-hour traffic or circling the block in a city center to look for parking.
Beyond this revisionist take on automobile autonomy, I see self-driving cars as opening up entirely new vistas. When parents can call a Google car and put their children in the back seat for a ride to soccer practice, that increases autonomy. When an elderly person who can no longer drive can call an autonomous vehicle for a lift to the supermarket or to the art museum, that increases autonomy. When all of this is affordable—so affordable that anyone can pay for it—it will have brought about a massive net increase in autonomy for all and an important increase in equity.
Yes, we will be dependent on autonomous cars, but we have always been dependent. Here, the dependency is actually replaced with something more reliable. The child always needs to get to soccer practice, whether a parent or neighbor or a Google car is providing the transportation.
As you can tell, when it comes to self-driving cars, I’m a starry-eyed optimist. The reality is that it won’t be an easy journey. Look at the negative publicity that Tesla got when a driver in Florida lost his life while his car was in autopilot. He trusted the system more than he should have, and Tesla got all of the blame. There were calls to outlaw the technology. And there surely will be further fatalities because of software imperfections and human error. These will be a tiny fraction of the number of lives that use of the technology saves, but it won’t matter—we will blame the machines.
We can also expect that reckless humans will cut self-driving cars off and jump lights—because they know that the cars are programmed to stop and give way. There will be street battles between man and machine.
The transition will be as traumatic as the battles between the horseless carriage (as the first cars were known) and the horses for supremacy of the roads. Of course the cars won, but we traded one set of problems and risks for another and one type of dependencies for another. And there will have to be a loss of autonomy for the drivers because we will ultimately have to take this away—they are too moody and dangerous. They will become the drivers in the driverless car.