"A computer can never do that"

Quote:

Originally Posted by gumboot (Post 6890706) And there's the issue. At least with a human driver you can feel like someone is to blame; there was a mistake and something went wrong. With a computer driving, you'd have death and injuring as a part of a working system. And therein lies the issue. I mean, how does this car detect other traffic dangers? Can it detect cyclists? Motorcycles? Debris on the road? Pedestrians? Does it detect a tyre blow-out or wobbly steering? One of the classic examples, for me, is pedestrian crossings in a big city. You'll always, always get people running across after the pedestrian crossing sign has gone read. Every human driver knows not to blast off the moment the light turns green. They can watch the footpath, determine if some idiot is going to run across the front of their car, and act accordingly. A computer can't do that. It will never be able to do that. Every single time the light went green at a major intersection a pedestrian would get flattened.

The post I quoted above was made in 2011. In retrospect, it is hilarious.



Here is where self-driving cars are now: When A Self-Driving Car Meets An Indecisive Cyclist



While still not perfect (although already significantly better than an average human driver), Google cars already do everything gumboot claimed "computer can never do". And read cyclist's hand signals too.

via JREF Forum http://ift.tt/1pKOwQf