This is the second article in a series revisiting some of Lawyers Weekly’s best stories over the years, seeing how issues have evolved since we last covered them, and how well our stories anticipated future developments.
In December 2014, we looked at the rapid development of autonomous vehicles—cars that drive themselves—and asked whether the legal profession was adequately prepared for a new technology with the potential to revolutionize almost every aspect of the way we get from place to place. Today we find that, in fits and starts, both the law and technology are moving forward.
The future of driverless cars somehow seems both closer and further away than it did four years ago. The car that shows up at your front door and whisks you off to your chosen location may take a little longer to arrive than optimists had predicted.
“I think we were laughing at how there was this common claim that the technology was ready if only the lawyers would get out of the way, and I said that was not at all accurate,” said Bryant Walker Smith, an assistant professor at the University of South Carolina School of Law and expert on the law of driverless vehicles. “I think the last four years have really been vindicating in that regard because I think there’s been a lot more focus on the reality of the technical challenges.”
But in a much more limited set of circumstances, driverless cars may be on the cusp of real-world deployment. General Motors has announced that it plans to launch a fleet of fully autonomous taxis in a small number of hand-chosen, urban environments in 2019. That’s the most ambitious publicly announced timetable, but multiple rivals are trying to pull off something similar.
So the question is perhaps more timely than ever—is the law ready for a car that drives you?
Moving in stereo
Different states have taken different approaches to the run-up. North Carolina has been comparatively proactive: last year it passed an industry-friendly law to regulate fully autonomous vehicles and amend existing motor vehicle laws to account for their presence on roads. While far from comprehensive, the law defines key terms, clarifies that a vehicle’s registered owner is responsible for any moving violations, and created a Fully Autonomous Vehicle Committee within the state’s Department of Transportation.
Conversely, South Carolina is one of several states that have declined to pass any major legislation related to autonomous vehicles. But Smith said that this inactivity doesn’t mean that South Carolina’s lawmakers are asleep at the wheel. (That would be ironic given that the state is a hub for developing AV technology.) Rather, those states have implicitly recognized that autonomous driving, or at least some forms of it, is already lawful in those states.
“All of these things are happening on different timelines and in different ways, in some ways with different perspectives,” Smith said. “The result is that we’re really learning a lot about how to think through these issues … the law is moving along in a messy, but ultimately useful way.”
These differing approaches may soon start to converge, however. The Uniform Law Commission’s committee on highly automated vehicles, on which Smith is a reporter, is currently drafting a uniform law that would cover the deployment of cars that are in complete control of vehicle functions. (States would then decide whether to adopt it.) A proposed federal law, the AV Start Act, was introduced in the U.S. Senate last year, although that bill appears to be parked in the garage for the foreseeable future.
And states have been willing to act in situations where existing law creates a clear impediment to the advancement of autonomous driving. Both North Carolina and South Carolina passed laws last year carving out an exception to their following-too-closely laws to legalize “platooning,” a system in which human-driven trucks travel in caravan. The trucks communicate with each other in order to automate braking, which allows the trucks to travel more closely together and thus save fuel.
Who’s gonna drive you home
Earlier this year, autonomous vehicles reached an unwanted but inevitable milestone when an Arizona woman became the first pedestrian killed by one, after she was struck by a self-driving Uber taxi. (For perspective, human-driven cars caused nearly 6,000 pedestrian deaths last year.) The tragedy exposed numerous flaws in Uber’s autonomous vehicle testing, which was promptly suspended. Smith said it was a crash that never should have happened—but also one that was easily analyzable under existing law, aided by the data captured by the car’s computers.
“There are 15 pedestrians killed on U.S. roads every day, and rarely is there the kind of evidence that protects the pedestrian’s interests, because when only one side is still alive, it’s easier to believe their side of the story,” Smith said. “I don’t want in any way to be macabre or dismissive of any traffic fatality, but the sad reality is that people who are struck by an autonomous vehicle at this point probably have a better chance of recovery than someone who’s struck by a human driver because so much attention is on these potential incidents.”
But the incident helps illustrate that we’re still a long way from cars that can operate anywhere, in any conditions. That’s part of the reason why the immediate rollout of autonomous vehicles will be limited to a very narrow set of conditions for which companies are confident in the performance of their systems.
Urban areas are particularly favorable terrain because vehicles can travel at slow speeds, within small areas, and plying the same roads repeatedly, in order to minimize the risk of injuries. So the first real-world driverless cars will be deployed in cities, aided by humans in remote operating centers who will be able to take over if the cars ever get confused.
A newer worry related to autonomous vehicles is that hackers could potentially hijack them and turn them into instruments of mayhem—although this overlooks the fact that hacking is difficult, whereas it’s actually quite easy to weaponize a conventional automobile. Smith said the vulnerability of autonomous systems is a real concern, but one that will be present in almost every part of our increasingly connected lives, and that would take a tremendous amount of effort on the part of bad actors.
“With cybersecurity, people who are unconcerned should be much more concerned, and people who are terrified should find many other things to terrify them,” Smith said.
Such anxieties could influence the way that autonomous cars are designed, however. Even once cars are fully capable of driving themselves, should humans have the power to take control in the event of an emergency? The answer might depend on whether the risks of possible malicious actors are deemed to outweigh the benefits of having a flesh-and-blood backup system onboard.
Follow David Donovan on Twitter @NCLWDonovan