4 ms·

I think this brings up an interesting philosophy for autonomous driving technology - is it feasible if it's "only as good as human driving"? I mean I've seen c

I think this brings up an interesting philosophy for autonomous driving technology - is it feasible if it's "only as good as human driving"? I mean I've seen comments that a human might not have seen the pedestrian. Is that a "defense" of driving AI? That's it's about as good as human driving? Seems to me the public won't accept driverless cars unless there is significant evidence that it's much better than human drivers - after all human drivers make a lot of mistakes and cause a lot of fatalities. I don't think any other claims of the advantages of AI driving could offset any negative publicity of injury or death. Just a thought in the context of this tragedy.

dfxm12 9y ago
I think experience is the best teacher. In a perfect world, Uber will look at the data, determine why this happened. The AI will learn and it won't happen again. That is compared to countless new human drivers entering the road every year. That's not a defense, or an excuse, but an explanation or reality. I know we don't live in a perfect world, but regardless, we can't change what happened here, so it's best to learn something from it.
- kbwt 9y ago
  > The AI will learn https://xkcd.com/1838/ https://xkcd.com/1838/
erobbins 9y ago
Undistracted, sober humans actually don't make a lot of mistakes. Distracted and impaired humans do, and when a mistake is made we can put them in jail. Who can we put in jail for software errors?