Vlad Tenev

Like both because it's hard, but mostly because you have all these like crank amateur mathematicians that are like, I've proved this big theorem.

but nobody's even going to pay attention to it.

Because to get a mathematician to concentrate and most of the time there's some mistake, but you have to really concentrate to find it.

But what formal verification results in is the machine can check the proof.

Yeah.

So you don't have to convince anyone.

If it compiles, it's guaranteed to be accurate.

I gave a TED talk about the future of AI.

And there's a lot of doomerism out there where people think nobody's going to have any jobs because the AI models are going to do everything that we can do, but they'll do it a thousand times faster and better.

I'm probably a little bit more optimistic for a couple of reasons.

One is that there's so many of these things that we don't understand.

Like we don't understand what's going on with the fundamental fabric of reality, which you can see with simple things like the double slit experiment.

We don't know what happened at the beginning of the universe.

We don't understand what happens with out of body experiences.

So that tells me with a fairly high degree of confidence that there's some aspects of what we consider intelligence that are quite possibly outside of your own brain.

It's probably not that, hey, if you understand just how the brain works and image it and duplicate it and copy it and move it into the cloud that you can replicate what's going on here.

which makes me think, well, there's probably some aspect of our intelligence that can't be replicated by just AI models getting better and better and faster and faster.

I think you also see it, you know, AI capability is growing.

but it's not like clearly overshooting human capability.

It's like we saw a big jump from generation three of the Transformers to generation four, but then it's kind of like at some level where it's always questionable whether it's smarter than humans or not.