In Machines We Trust: Exploring AI's evolution (Live)
MIT Technology Review Narrated
00:00
I Don't Like the Terminology of Artificial Intelligence
I don't like the term AGI because I don't think human intelligence in general. We should have architectures that allow them to reason and I'm proposing one, which may or may be right. And attempts to reach human level intelligence by just scaling or just making or current learning algorithms better is not going to work. There's no process by which an answer is being elaborated with an unbounded number of steps. So what about reasoning? What about planning? It's a vision where I say we should get machines to learn models of the world.
Transcript
Play full episode
