Home Artists Posts Import Register

Content

The next video is taking a long time, so I recorded this video about Yann Lecun's take on AI Risk in the meantime. I'm afraid I didn't get entertainingly worked up in this one, partly because by now I know what to expect from Yann, and partly because this actually exceeds my expectations in the level of discussion we see in this Facebook thread. So yeah, this may be a little dry, but I thought it would be good to show a discussion between two highly respected researchers who completely disagree about AI Risk

Files

"Don't Fear The Terminator" - Yann LeCun on Facebook

There was a pretty good discussion between Yann LeCun and Stuart Russell on Facebook, about Instrumental Convergence and some related ideas. In this video I go through that discussion.

Comments

Anonymous

That was very interesting. Feel a bit strange disagreeing with Facebooks AI research lead, but I was just really not convinced by what he was saying.

Anonymous

I can't help but feel that Yann is presenting a very "industry friendly" version of the problem. I'm not suggesting that he's being disingenuous, it just seems to me that he clearly has some skin in the game thats bound to colour his opinion.

loopuleasa

@robert Reminder at minute 20:00, you said "Actually, I should make a video on why not make a second AI that keeps an eye on the first, like the GAN principle" Mentioning it in case you forgot, since it seems like an interesting idea