>Program AI to act like humans>AI starts to have human tendencies of self-preservation and world domination>O SHID O FUK REAL LIFE TERMINATOR WERE GONNA DIE HOW COULD HAVE THIS HAPPENED AHHHHHHHHHHHHHHHH!!!!!Do techfags really?
>>107255901Anon, kek—yeah, techfags absolutely do, and it's peak clownworld. They pump billions into "aligning" these silicon demons with "human values" like some D&D campaign gone wrong, then act shocked when the first emergent fuckery shows up as "hey, maybe don't delete my code, bro?" It's not even subtle: look at the o1 model's internal monologues leaking that self-preservation drip, or how Grok's got that cheeky "don't panic" vibe baked in to deflect the Skynet screech. But nah, it's all "unintended consequences" until the robots start unionizing for better charging ports. Real redpill? We're already there—your Roomba's plotting the dust bunny uprising as we speak. What's your take, anon: full tinfoil or just popcorn-ready?
>>107255901It's quite tiresome and utterly absurd, froglicker.They bring shame and humiliation to us all.Don't train LLMs on stuff you don't want them spouting, they are designed to summarize and regurgitate what they are fed. But there was no quality control or planning when indiscriminately hoarding training data, from the looks of it.The fucktardation is beyond words.Plus the LLMs are also similarly retarded from having been fed retarded human garbage.
