Honestly, I think giving robots AI is kinda what'll be necessary to create a general AI. I know that's not what they're talking about here, but I speculate that part of the reason why AI hallucinates so much is that it has no "reality anchor". It has no way of creating real-world experiences for itself, so fact and fiction are just funny words.

*Yes, I'm aware that LLMs don't really understand words, or at least they don't appear to, however when it comes to attempting to create true intelligence I think it'll need some form of anchor to keep it from hallucinating like LLMs do.