Anca Dragan on why Asimov's three laws of robotics need updating
The Robot Brains Podcast
00:00
Is There a Future for AI?
Soprovise learning is a way of training computer systems to be aware of their limitations. It's not about making the system do things automatically, but rather giving it an awareness of what its limits are. The problem with high capacity models such as those used in cars and robots is that they can pick up on spurious correlations between different data sets. And if you're not careful, one model could end up being the wrong one.
Transcript
Play full episode
