I have yet to see an LLM with hands, feet, or eyeballs.

Currently, LLMs require hooks and active engagement with humans to ‘do’ anything. Including learn.