The overwhelming majority of motor vehicle accidents in California and around the country involve a driver who made some kind of human error, according to data from the National Highway Traffic Safety Administration (NHTSA). It is widely believed that fully autonomous vehicles will virtually eliminate road deaths and injuries by taking human beings out of the driving equation, but some experts are not convinced by this argument. According to researchers from the American Automobile Association, vehicles controlled by algorithms rather than people will only cut traffic accidents by about a third.

After studying more than 5,000 accidents from NHTSA’s National Motor Vehicle Crash Causation Survey, researchers from AAA’s Insurance Institute for Highway Safety concluded that only about one in three of them were caused by driver mistakes that autonomous technology would have prevented. To improve safety further, the researchers say that self-driving cars would have to be programmed to prioritize safety over driver preferences like convenience and speed.

These shortcomings were highlighted in March 2018 when an autonomous SUV being tested by the ride-sharing firm Uber struck and killed a pedestrian in Tempe. A study of the information stored on the SUV’s data recorder revealed that the autonomous system noticed the pedestrian, but it did not anticipate that she would attempt to cross the street. When she did, the Uber vehicle failed to take the appropriate evasive action.

Self-driving cars could also change the way that motor vehicle accident lawsuits are litigated. Instead of taking legal action against negligent drivers or their insurance companies, personal injury attorneys with experience in this area could file lawsuits against companies that manufacture vehicles or their autonomous systems. While the defendants could be different, attorneys may still seek to recover damages to cover their client’s medical expenses, property damage and lost income.