Corey Knowles

I don't know.

I'm not sure.

Where we go from there, I'm just kind of remaining confident in the idea that, you know, the researchers that we talked to and the people we've been chatting with along the way who are who are building this stuff, they seem to be really confident that we've still got a lot of road ahead and that it's going to improve.

Yeah.

And pretty quickly, we as humans have proven that we will absolutely trust anything that

pretty quick and easy.

We do that in every aspect of human life.

And I don't know why I thought humans would be more discerning about AI copy as far as when they're working with it and using it and making sure to give it a good overview before they're, you know, running out and publishing things or whatever the case is.

But boy, we really don't as people.

We've kind of just accepted it.

Unless you're in your specific field where you have the knowledge, where you are aware of those nuances, it can be really convincingly wrong if you have a cursory knowledge of something.

Absolutely.

As opposed to being an expert in your field.

I find that people who are with that said...

People who are really good at a thing tend to be better at it when they understand how to use AI in a smart capacity, like where it can be strong and understanding where it is not.

And it's one of the reasons I love to bounce between models and companies.

But the truth is, when I'm doing my work, I'm doing most of that in ChatGPT, just largely because I know where it's going to fall short.

I, you know, it's what I've used for, if it's an important task, I'm going to use the tool that I know what I need to watch in, as opposed to one where I'm maybe less sure.

Like, I mean, it's probably just as good.

You know, the truth is, all of these models are good now by comparison to, when you look back at what we were working with two and a half, three years ago, it's a...