NEW YORK — Picture it. You slip into your car, recline and start reading the day's news on your smartphone. The vehicle accelerates, smoothly navigates traffic and seamlessly merges onto the freeway, without your lifting a finger.
All around you, other autonomous cars zoom by. Each operates safely at high speeds, halving your commute time. Each drives with precision, eliminating congestion and conserving fuel. Each respects pedestrians, avoids collisions and always takes the most efficient route. Your car drops you at the office, then parks itself in a lot on the outskirts of town, where it awaits your summons at quitting time.
Such a glorious commute may be decades away. But the era of autonomous cars is fast approaching. Last week, Audi and Toyota both made headlines with advances in automated driving. Google has been a much-publicized pioneer in the field. Automakers General Motors, Daimler and Nissan, among many others, have plans to automate their products to varying degrees.
And no wonder. Autonomous cars could create lucrative new businesses, spur welcome advances in public planning, vastly improve our quality of life, mitigate human error on the roads, and, not least, reduce the more than 30,000 annual driving fatalities in the United States alone. There will also be drawbacks and unintended consequences. That's why policy makers need to start planning for a future that seems to be arriving faster than we ever expected.
Three issues in particular require scrutiny.
First, we'll need a new legal and regulatory framework. State and federal driving laws obviously weren't written with this technology in mind, but as Bryant Walker Smith of Stanford University's Center for Internet and Society has persuasively argued, automated vehicles are almost certainly already legal throughout the U.S.
California, Florida and Nevada have already begun regulating them. Such laws will start important conversations: They'll acclimate regulators, transportation authorities and law enforcement to the new technology, and offer manufacturers some degree of predictability. State lawmakers should also develop common legal and technical definitions; words such as "driverless" and "automated" are often used blithely yet can mean very different things. And states should consider how they'll need to update their licensing and registration rules and revise traffic laws — for instance, by reducing following- distance requirements when appropriate — to better accommodate autonomous technology.
Federal legal clarity is also essential. The National Highway Traffic Safety Administration, which is planning to study the issue, should be transparent with the public and the auto industry about any plans to update its rulebook, and when it intends to act.
In doing so, it should remember that there is a wide spectrum of automated technology that runs from anti-lock brakes to Google's driverless car and beyond, so regulation must be flexible. And it should take special care not to inhibit innovation in a quickly evolving field. This is easier said than done, but one approach would be to focus more on conceptual standards — such as requiring that autonomous cars not exceed a certain number of accidents per million miles — rather than mandating specific safety features.
Next, policymakers will have to devise clear rules for determining liability. Should the driver be held responsible for his vehicle? The manufacturer? Perhaps a third party that designed an app for a car that malfunctioned? Resolving such questions will be incredibly messy. A smart first step would be to require event data recorders — "black boxes" that contain information about a car's operation before a crash — in all autonomous cars, with adequate privacy safeguards.
Congress should also consider the model offered by the National Vaccine Injury Compensation Program. Autonomous vehicles may in the future be judged an essential social good. But, as with vaccines, the potential for rare but extremely costly lawsuits could inhibit manufacturers from meeting demand. To ease such concerns, Congress could direct claims against autonomous-car manufacturers to a special federal court outside the normal tort system, and establish a fund to compensate accident victims.
Finally, officials at all levels must manage expectations, and plan realistically. Autonomous cars won't be perfect, and they will surely cause havoc in many cases. Their benefits may also be exaggerated in the public mind. In the best long-term case, a commute like the one described above — faster, safer, more productive — could become a reality. But not soon. In fact, as Smith argues, high demand for autonomous cars could in the medium term prevent heralded reductions in congestion, emissions and sprawl.
Because one near-certainty of the driverless era is that more people will want to drive. That will have wide-ranging policy implications, including the need to modernize the way we finance transportation infrastructure. That, in turn, means more user charges, such as congestion pricing and variable tolls, and especially a vehicle-miles-traveled tax to ensure that drivers bear the cost of road upkeep.
Ceding our roads to computers will be a complicated and chaotic process. It will involve all kinds of strange moral considerations and trade-offs. It could transform our physical environment, our economy and quite possibly our entire way of life. The technology is unquestionably revolutionary. Just don't expect it to be a joyride.
