Autonomous driving and the economics of blame
Autonomous driving and autonomous vehicles appear on 314 recent papers, 270 of them in AI journals about perception and control, and only 29 carry a liability, insurance, regulation, acceptance or ethics keyword, with liability on 2. Driverless cars are now on public roads. When one crashes, who pays, and does the answer shape what gets deployed?
What prompted this
Since October 2025, 314 papers carry an autonomous-driving, autonomous-vehicle, self-driving or automated-vehicle keyword (270 in AI journals, 31 in operations research): "autonomous driving" on 152, "autonomous vehicles" on 64, "autonomous driving perception" on 45, "connected and automated vehicles" on 15, "autonomous driving safety" on 10. Only 29 of the 314 also carry a liability, insurance, accident, regulation, acceptance, trust, ethics, law or policy keyword, and "liability" appears on 2 papers.
Questions
- In crashes involving driverless vehicles so far, how was fault assigned between the operator, the manufacturer and other road users, and how did insurers settle?
- Do strict manufacturer liability regimes slow deployment or improve safety, compared with regimes that treat the vehicle like a driver?
- How do crash rates per mile of driverless fleets compare with human drivers on the same roads, once reporting differences are corrected?
Data that could help
Crash reports filed with regulators (the US standing general order data, state DMV reports), court and settlement records, and fleet mileage disclosures.
Next step
Compile the crash reports of driverless fleets since 2021 with fault determinations and compare injury rates per mile with human-driven ride-hailing in the same cities.