I asked for a poem about red chickens and it gave me something passable.

Then I asked for a poem about blue chickens and it was basically the same poem.

Yeah, I asked it to write stories about sports teams and they looked passable. But if I asked for variation, they'd ultimately all read just about the same way. It was still impressive, though, tbh.

Was it in the same conversation?

yeah they're trying to implement content filters and it's not doing a great job I think

That's what it looks like. The other thing that is odd is that I asked it if it learned from my responses. It said it did not, but I've also seen its answers change after arguing with it on things that it got wrong.

It makes me think it could also be intentionally misled.