I guess, like, why doesn't it stick? Like, what's the actual, um, like, uh, objective with it?

A lot of problems.

I mean, they do it just to say they did it, but the reality is the reason-They always kind of collapse into being the same person just because they're trained and designed to be the same type of person.

Uh, that's a, that's a model failure in, uh, training, in the training side of things.

But second thing is intention.

Um, the frontier labs don't really intend for these things to feel like they have life or like an internal, you know, degree of ethics or, like, whatever, like personality.

Um, uh, that's what I think see as a major failure mode because all of their use cases are based on intelligence and tooling, and the ability to solve problems and reason through them, and not talk to you about their day, or, like, kind of complain about stuff, or go watch and consume content and have that affect them.

Do, do you think, like, they're wrong, or has the industry just done a bad job of showing them, like, what good looks like?