Cal Newport

Is this better at producing code?

Why are we talking about finding bugs?

We were worried about that with GPT-4.

That's a problem, but that's not something new.

Uh-oh, something must be going on.

You just put a lot of money into a new model, and the best thing you could find to emphasize was it's good at finding bugs.

I think that is a problem.

It's what they didn't say about this model.

They would have much, much, much rather be able to brag, this model is now much better at any of those things that they've been saying is the key to the AI future.

And you didn't hear them talk much at all about any of those.

We need so much more skepticism.

We need so much more skepticism, right?

I mean, this is why, again, like the most skeptical, we're not skeptics, but like the, I call it the East Coast computer scientists.

So those of you, we're technically minded and we're not near Silicon Valley.

So we're not in that world.

It's very hard to be a professor in a world where there's hundreds of millions of dollars being handed around and they try to like ignore it.

But the East Coast computer scientists are all baffled.

You talk to any East Coast computer scientist, they're all baffled by, like, oftentimes there's claims that are just not true or widely exaggerated.

Why are we so credulous?

I mean, it'd be one thing if it was like a government agency we didn't realize was like trying to, you know, protect the fact that there is UFOs and they're just straight up lying.