BlushedPotatoPlayers@sopuli.xyz to Technology@lemmy.worldEnglish · 10 months agoAI chatbots tend to choose violence and nuclear strikes in wargameswww.newscientist.comexternal-linkmessage-square52fedilinkarrow-up1173arrow-down121file-textcross-posted to: worldnews@lemmit.onlinenottheonion@lemmy.world
arrow-up1152arrow-down1external-linkAI chatbots tend to choose violence and nuclear strikes in wargameswww.newscientist.comBlushedPotatoPlayers@sopuli.xyz to Technology@lemmy.worldEnglish · 10 months agomessage-square52fedilinkfile-textcross-posted to: worldnews@lemmit.onlinenottheonion@lemmy.world
minus-squareEven_Adder@lemmy.dbzer0.comlinkfedilinkEnglisharrow-up2arrow-down2·edit-210 months agoIf that’s really how they work, it wouldn’t explain these: https://notes.aimodels.fyi/researchers-discover-emergent-linear-strucutres-llm-truth/ https://notes.aimodels.fyi/self-rag-improving-the-factual-accuracy-of-large-language-models-through-self-reflection/ https://adamkarvonen.github.io/machine_learning/2024/01/03/chess-world-models.html https://poke-llm-on.github.io/ https://arxiv.org/abs/2310.02207
minus-squareMNByChoice@midwest.sociallinkfedilinkEnglisharrow-up1·10 months agoI will read those, but I bet “accidentally good enough to convince many people.” still applies. A lot of things from LLM look good to nonexperts, but are full of crap.
If that’s really how they work, it wouldn’t explain these:
https://notes.aimodels.fyi/researchers-discover-emergent-linear-strucutres-llm-truth/
https://notes.aimodels.fyi/self-rag-improving-the-factual-accuracy-of-large-language-models-through-self-reflection/
https://adamkarvonen.github.io/machine_learning/2024/01/03/chess-world-models.html
https://poke-llm-on.github.io/
https://arxiv.org/abs/2310.02207
I will read those, but I bet “accidentally good enough to convince many people.” still applies.
A lot of things from LLM look good to nonexperts, but are full of crap.