Itamar Friedman: belief

Transcript context

…when the agents take over exploration. So what do you think is the most interesting unsolved question in, in ai? Like, what would you re, what's an open question that you think, man, somebody should solve that? Okay, so here I am going to go to the Yes obvious answer. That's a AI alignment. Mm-hmm. Like, it's, it's a technical question. It's it's a philosophy question, et cetera. It's, it's, it's not easy. Like it raises so many question even about ourself as as human or we, like, I saw one tweet by someone that I'm thinking about like for a few years he wrote are we actually like LLMs, like in essence? So, so I think like we're trying to look into those LLMs for years. Like there, there was, like in 2014 there was already in the C N N, there was a few works. Trying to visualize what, what are the, the feature detection, the feature, like what are the feature with the hidden layers that you see, like we're trying to work on it for years, lately, like a really long time ago, like five years, days ago or so, like, we saw work by open ai, like trying to turn, look on on different parts of Dell LM and trying to provide a natural language description for them. So I think like this is very important. Very interesting tech-wise, philosophy wise, et cetera, that that's like, I think need to be explored more. And just one takeaway for all the listeners, like what's one message you want everyone to remember about ai? I, I…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.