I find this less annoying to read if I have Claude rewrite it with the following prompt:
rewrite this lesswrong post to be less human-chauvinistic
don’t change the overall thrust of it at all, keep the same points but just avoid saying things like “there aren’t live mental elements underlying the LLM text” (super debatable)
in other words rewrite it in line with the spirit of this top comment:
Even if we assume or it becomes true that LLMs are genuinely minds and all that, it still seems similarly bad to use them like this. LLM-generated text is not your testimony, regardless of whether or not it is ‘testimony’.
I find this less annoying to read if I have Claude rewrite it with the following prompt:
rewrite this lesswrong post to be less human-chauvinistic
don’t change the overall thrust of it at all, keep the same points but just avoid saying things like “there aren’t live mental elements underlying the LLM text” (super debatable)
in other words rewrite it in line with the spirit of this top comment:
lmk if that’s unclear at all