[godshatter]

I wonder what would happen if LMs were built a bit at a time by:

  - add in some smallish portion of the data set
  - have LM trainers (actual humans) interact with it and provide feedback about where the LM is factually incorrect and provide it additional information as to why
  - add those chat logs into the remaining data set
  - rinse and repeat until the LM is an LLM

Would they be any more reliable in terms of hallucinations and factual correctness?

This would replicate to some extent how people learn things. Probably would really slow things down (not scale) and the trainers would need to be subject matter experts and not just random people on the net say whatever they want to say to it as it develops or it will just spiral out of control.