Sure, but the LLM needs to prove that it can make inferences as well as or better than a human, in order to be useful. Aside from that, it's not human, so there

Sure, but the LLM needs to prove that it can make inferences as well as or better than a human, in order to be useful. Aside from that, it's not human, so there's no need to be fair - it should do what we tell it, not decide on its own.