Phil

I agree.

Yeah.

It all boils down to expropriating property.

A lot of what you're saying is context dependent and it's also model dependent, right?

Like, so I just asked my AI, I was like, hey, what are the 10 best white basketball players of all time?

And he came back with no, there was no, no hemming and hawing.

There was no extra.

Was Dirk on there?

What?

Was Dirk on there?

No, he's got Larry Bird.

Yeah, Dirk Nowitz, Steve Nash, Pete Maverick.

So like he, there was no, no, I didn't have to cajole him.

He said it depends on the context and how you define racism.

And then he goes for the breakdown.

The charitable reading, the honest problem, the asymmetry question so he recognizes it.

It really does depend on the model you're using and how you- Previous prompts.

Yeah, a little bit.

Like I'm using a program called Open Claw and it has there's certain things that I've told it to remember to do that I don't want bullshit.

I don't want, you know, these kind of things that he knows, like he knows that I'm right leaning, you know, so it depends on how you it's like it is a tool and it depends on how you use the tool.