I agree with the problem (which is rumoured to be one factor in the Trump regime’s unhinged behaviour), but I doubt that an add-on to improve LLM output will do any more than polish the turd. And how would such an add-on be created? More LLMs would only pile the shit higher. Even humans don’t seem capable of “revising” LLM slop into anything useful.
ETA: “anything useful” depends on one’s use for it. Clearly (if the rumours are true) it’s being very useful to the Trump regime. They can do anything they like and trot out an LLM argument for it, and do the same to answer all possible reactions and all subsequent events. What people seem incapable of doing is “revising” LLM output into something possessing epistemic virtue.
I agree with the problem (which is rumoured to be one factor in the Trump regime’s unhinged behaviour), but I doubt that an add-on to improve LLM output will do any more than polish the turd. And how would such an add-on be created? More LLMs would only pile the shit higher. Even humans don’t seem capable of “revising” LLM slop into
anything useful.ETA: “anything useful” depends on one’s use for it. Clearly (if the rumours are true) it’s being very useful to the Trump regime. They can do anything they like and trot out an LLM argument for it, and do the same to answer all possible reactions and all subsequent events. What people seem incapable of doing is “revising” LLM output into something possessing epistemic virtue.