They definitely hallucinate a lot too. But they also seem to do things genuinely like reasoning, e.g. they seem to contain and use "cognitive" world models, and seem to be capable of learning fully general algorithms.

Beyond that, many intellectual capabilities are also useful for successful bullshitting, so even if we can confidently say that that's all they do (in some sense), that doesn't mean they don't also do something-like-human-reasoning etc.