I don’t know what to tell you. We call it “reinforcement learning”, but it’s obviously nothing like actual human learning. The difference between rlhf and actual learning is much bigger than between guiding and training.