Because we already have artificial (man-made) general (contrast with domain specific) intelligence (algorithmic problem solvers). A.G.I.

If ChatGPT is not AGI, somebody has moved goalposts.

Both "general" and "intelligence" are _at least_ easily arguable without moving any goal posts, not that goal posts have ever been well established in the first place.

I think a baseline requirement would be that it… thinks. That’s not a thing LLMs do.

That’s an odd claim, given that we have so-called thinking models. Is there a specific way you have in mind in which LLMs are not thinking processes?

I can call my cat an elephant

it doesn't make him one