The media loves to frame matters of technology in quite dystopian terms, so it's perhaps not surprising that coverage of autonomous vehicles has tended to focus on both the safety implications of the technology and the number of driving jobs that might be lost. It's a narrative that seems to demand perfection from the technology before it can be rolled out en-masse on public roads.
Such thinking has distinct dangers of its own, with a recent from the RAND Corporation highlighting how delaying the launch of autonomous technology until perfection is achieved will cost many thousands of lives per year.
The Benefits of Imperfection
The report argues that even a 10% improvement on human drivers could save thousands of deaths on our roads, thus rendering it morally questionable whether it's wise to demand technology that is 90% and above better than human drivers.
The figures were arrived at by examining hundreds of possible futures and the changing safety requirements for autonomous vehicle introduction, with estimated road fatalities extrapolated for each potential scenario.
"Our work suggests that it is sensible to allow autonomous vehicles on America's roads when they are judged to be just moderately safer than having a person behind the wheel," the authors say. "If we wait until these vehicles are nearly perfect, our research suggests the cost will be many thousands of needless vehicle crash deaths caused by human mistakes. It's the very definition of perfect being the enemy of good."
Whilst autonomous technology offers a wide range of advantages, from greater convenience to a lower environmental impact, the safety benefits are perhaps the most persuasive. It's estimated that over 90% of accidents are the result of driver-related errors, and it doesn't take a huge improvement in that performance to yield significant reductions in injuries and fatalities on our roads.
Psychologically, however, the authors accept that even though minor improvements in safety performance would yield significant benefits, society is unlikely to accept autonomous vehicles on our roads until performance increases significantly.
"We may be less tolerant of mistakes made by machines than of mistakes made by people," the authors say. "But if we can accept that early self-driving cars will make some mistakes — but fewer than human drivers — developers can use early deployment to more rapidly improve self-driving technology, even as their vehicles save lives."
The authors hope that the study will better inform policy in this area, especially as policymakers weigh up the risks and benefits of the technology. With an estimated 35,000 deaths and over 2.4 million injuries on American roads per year, it's a debate that needs to conducted in a rational rather than emotional way.