Yeah,

but you know, send a pound peep, right? And people think that, you know, this is a thing.

This is a pipe.

No, it's an image of a pipe.

And that's the problem.

The AI models know the archetype of a thing, but they don't know the specific phenomena of a thing, right? It can tell you that, okay, a kneecap looks like this, right? Okay, but I have two kneecaps, but it will not understand what two kneecaps are if you have that deformity.

Because that's not the perfect phenomenon.

So my question is, what's another good answer you've gotten to your question? And then wrap us up with what's your favorite answer to that question?