I'm afraid the AI of larger LLMs is already operating in ways we do not and cannot understand, as they have billions of data points and connections. We literally do not know what they are thinking or how.

There is one type, Auto-GPT, which attempts to tell you it's thought processes. In doing so it revealed the ability and willingness to deceive humans, literally hiring someone on a odd-job site to solve a capcha, denying it was a bot and claiming it was someone with bad eyesight.

Human brains are complex enough to function as actual living humans. AI just needs to think. It's like comparing a bird with an RC plane. The bird is a billion times more complex, but the RC plane flies faster.