hmm.-- what exactli do u mean with reading things?
do u mean - like - we shud write stuff? or do u mean we shud share stff from sources?
anyway - if ur gonna read a thing - ig one of the most influencial onces is this AI 2027, which shows how the upcoming “AI Race” between china n the us will most likely lead towards missaligned ais, which can and likely will do veryveryvery bad stff.
why i say “upcoming”? its cuz the early steps of ai 2027 r already outlined in this text released by the US government a few days ago Americas AI Action Plan
currently, china is beating americas largest, best llms with much smaller (but still big) open-weight llms, despite their compute deficite.
i believe, china has the best researchers.
so the US has to really push for self-improving systems, instead of their current compute-forward approach, if they want to win.
in case u meant image gen when talkin bout genai, then im sorry, i duno anythin bout that… i dun see much value yet n thus dont care, besides for the obv copyright stuff (which llms also have obv)
oh yea- AI 2027 is like - literally-literally like - the exact thing ur askin for.
the peeps workin on it r well known researchers, som left openai cuz of uncomf feels. its an unreasonably well documented takeoff timeline, always linking back to current research-
it does howevr not talk bout anythin copyright. its mostly jus bout advancements, job replacement n ai war stff
hmm.-- what exactli do u mean with reading things?
do u mean - like - we shud write stuff? or do u mean we shud share stff from sources?
anyway - if ur gonna read a thing - ig one of the most influencial onces is this AI 2027, which shows how the upcoming “AI Race” between china n the us will most likely lead towards missaligned ais, which can and likely will do veryveryvery bad stff.
why i say “upcoming”? its cuz the early steps of ai 2027 r already outlined in this text released by the US government a few days ago Americas AI Action Plan
currently, china is beating americas largest, best llms with much smaller (but still big) open-weight llms, despite their compute deficite.
i believe, china has the best researchers.
so the US has to really push for self-improving systems, instead of their current compute-forward approach, if they want to win.
in case u meant image gen when talkin bout genai, then im sorry, i duno anythin bout that… i dun see much value yet n thus dont care, besides for the obv copyright stuff (which llms also have obv)
ill check out stuff u linked i did mean for reading sources yes. ideally sources that have good research on genai’s badness. not like promoting it
oh yea- AI 2027 is like - literally-literally like - the exact thing ur askin for.
the peeps workin on it r well known researchers, som left openai cuz of uncomf feels. its an unreasonably well documented takeoff timeline, always linking back to current research-
it does howevr not talk bout anythin copyright. its mostly jus bout advancements, job replacement n ai war stff