Thread with 3 posts
jump to expanded post
one problem with the “some guy” model of LLM chatbots (https://social.noyu.me/@hikari/statuses/01JQC19YTPBN5V5GS7FCCB38Z3) is that a human is limited in how many people they can be the friend of, but ai models can easily have millions of users. i don't think we have a precedent for how much of an intellectual monoculture this could produce
we've definitely had hugely influential singular written sources before. wikipedia for example is too influential, simply because it's so accessible, but this could happen for any kind of media…
this was a different risk because you still had to process the information yourself
also, no singular written source could ever have the sheer breadth that something like chatgpt can. you really can just ask chatgpt anything and it'll have something resembling a reasonable answer. you really could just completely delegate your decision-making to it…