Ran into this talk by Demis Hassabis, thanks to a pointer from martin shervington just now. Here, Hassabis shows how…

Ran into this talk by Demis Hassabis, thanks to a pointer from martin shervington just now. Here, Hassabis shows how machine learning learns various classic Atari games, mastering them and absolutely devastating any competitor. As Hassabis puts it: 

“It sort of ruthlessly exploits the weakness in the system that it’s found. This is all automatic.”

That’s not why I’m sharing this though. The key place to concentrate is actually at minute 2:54 (click that number to jump straight to it), where he outlines three levels of information processing in the brain, which he notes are also tiers of work that is, and still needs to be, done in the field of artificial intelligence: 

* Perceptual

* Conceptual

* Symbolic

Scroll to Top