But we have a lot more sensory input and context to verify all of that. If you kept training LLMs with all that data, it would be interesting to see what the r

But we have a lot more sensory input and context to verify all of that. If you kept training LLMs with all that data, it would be interesting to see what the results would be.