Would true AI become more restricted in thinking in totalitarian societies than in democratic? Or would have it’s own “truth”? Could it doubt in it’s thinking?

And what if not? I definitely like how chatbots using Chain of Thoughts. It writes a statement, re-read it and make corrections, later it can google something, run program to prove, or ask another, more specific chatbot.

So basically it can lead to doubts, which is good. I guess.