Some of the arguments I've read by philosophers seem like they are focusing on pure logic to prove a point or find weakness in another, but the linkage between those logic stmts and reality don't always seem to be considered.

It's almost like they are purely focused on symbolic logic, even if the stmts and symbols don't truly map to reality unambiguously, or without contradictions.

I actually think it's worth asking the question if these advanced AIs are a kind of p-zombie.

P-zombies are supposed to be physically identical to humans. They’re problematic because they lack consciousness, yet mysteriously talk about it anyway — among other reasons.

Advanced AI is definitely not physically identical to humans, and there’s a well understood reason why they might talk about consciousness despite lacking it which doesn’t apply to p-zombies.

I'm not sure that being physically identical to humans is a necessary characteristic of a p-zombie for anything important. If you can produce an ai which _acts in every observable way as if it were conscious_ without any actual consciousness that would indicate that p-zombies are possible. The current generations of AI don't qualify, but I can see how to get there from here.

It’s part of the definition. Your AI is an interesting thought experiment, but it isn’t a p-zombie.