Ben Gilbert: evaluation

Transcript context

…It's an unanswerable question. It is because there's the last 2% of use cases, the tail use cases. It's unknowable what it will take to get past that last 2%. There's this pretty interesting philosophical question, which is, how safe does a robot have to be? In the US, I think there are 40,000 deaths as a result of car accidents. Let's say that robot cars are 10 times safer. I think that highway accidents are one of the top two or three causes of death in the United States, period, so something 10 times safer. Yeah. If you're 10 times safer, fast forward 25 years from now, who knows when it will be, 4000 deaths a year, so a little more than 10 a day. If you have four companies that are responsible for the marketplace, five companies, and there are 10 deaths a day, a good day is, hey, we only had one fatality. That's a good day. I can't imagine that. There's this, well, does it have to be 10 times better? I don't think it's good enough. It has to be 100 times better. Maybe that's not good enough. From a societal standpoint, of course, if it's 100 times better, we should go forward with it. But that would mean there are 400 fatalities a year, one every single day. I don't know how society would deal with it. Our society is very forgiving, but they understand humans are human, and humans make mistakes.…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.