Daily reminder that, as far as we know, the hedonic treadmill is not a law of the universe¹, evolution didn't optimise valence strongly in any direction, and there's <$10m a year going into making humans superhappy
¹: indeed we have good evidence that it isn't
One reason to not spend overly much time lawyering the meaning of words to minimize LLM’s capabilities is that you should not want to redefine thinking such that many humans have never thought.