RogerDearnaley comments on Human-like metacognitive skills will reduce LLM slop and aid alignment and capabilities