In a recent conversation I had with my mother about autonomous vehicles, she established that she'd never feel safe inside of a car that lacked a real driver. Her reasons are shared by many: The lack of control means higher stress, and of course the feeling of knowing that like all computers, the risk of an issue arising at an inopportune time (eg: while in traffic) is all too real.
While it might not come as a surprise to everyone, it's been proven before that Google's autonomous vehicles are actually safer on the road than those driven by actual people - and now, that proof has been furthered, after a slew of tests in California and Nevada.
Google's Chris Urmson states that "Our car is driving more smoothly and more safely than our trained professional drivers" - overall, the cars experience much less near-accident states than regular vehicles.
At some tasks, computers are just going to be better than humans - almost perfect in some cases, and without variation. From that standpoint, it makes sense to believe that these cars are a lot safer. People make mistakes, while it's rare to see a computer suffer one. There are of course other risks when it comes to a computer operating anything, but Google's confident that if we were all sitting in autonomous vehicles, road safety would improve dramatically.
It's hard to disagree there. At this point, the real hurdle isn't making these things work, it's simply getting them legal and in the hands of consumers.