I think the suggestion is that keeping track of how much current LLMs reinforce cranky beliefs will help you not use the same level of reinforcement from LLMs as evidence for your future beliefs that you may not realise are cranky.
I think the suggestion is that keeping track of how much current LLMs reinforce cranky beliefs will help you not use the same level of reinforcement from LLMs as evidence for your future beliefs that you may not realise are cranky.