An estimated 37,000-40,000 Americans die in car accidents every year. We now have substantial, real-world evidence—from Waymo and joint analysis with Swiss Re—that driverless operations may be safer than simulated human driving within their current operating environments. Recent data shows that over 220 million miles, Waymo vehicles—in Los Angeles, San Francisco, Phoenix, Austin and Atlanta—have 94% fewer serious injuries, 82% fewer air bag deployments, and 93% fewer pedestrian injuries. The evidence is not fully conclusive, but it is unusually clear, substantial evidence.
So with thousands of lives every year in the balance who is against autonomous vehicles (AVs)? Trial lawyers. Amazingly, trial lawyers saw the writing on the wall early on and have been pursuing AVs for nearly a decade! The American Association for Justice, a forum for trial lawyers, was strongly opposed to the AV law (see also reports here). (They were joined by Democrats who are concerned about jobs and want heavy trucks out).
Trial attorneys earn huge sums in typical auto accident lawsuits–Annual US auto insurance payouts (liability + PIP/MedPay) are on the order of $180–220B and trial attorneys are eager to retain the right to sue automakers for product liability. In my opinion, product liability does not work as a safety device in this industry. Instead, the solution is simple. Every car should be required to have insurance, regardless of who the driver is. Indeed, Waymo vehicles are already insured for $5 million in liability per vehicle, much higher levels than most human drivers.
The UK’s Automated and Electric Vehicles Act 2018 basically does this–single car insurance covers the car whether a human or an automated system is driving; the victim is directly compensated by the insurance, there is no need to establish a product feature; the insurer then becomes liable against the manufacturer if the software was faulty. Victims are paid immediately, manufacturers cover the costs of their disabilities through discovery and premiums, and high operating costs (ie attorney fees!) and messy producer-versus-victim lawsuits are replaced by insurance-producer negotiations between successful repeat players.
The best thing about this plan is that the insurance is almost guaranteed better there is abuse: airlines generate data that makes the experience measurement accurate, so insurers become continuous regulators of security, while the case brings a noisy, forgotten, lottery signal depending on the unimportant safety factors of the judge and the place.
We have the best data on Waymo, the Tesla data is murkier but note how well this works with the insurance plan. Let the insurance providers decide how much to charge Tesla robotaxis and FSD drivers-they will cover the interior much better than the tort lawyers. In short, insurance works well for accident victims but not for trial lawyers. Of course, if the trial lawyers have their way accident victims will continue to be buried in invisible graves.
Hat tip: Andy Hall and Jon Slotkin.
