pandasiusfilet@feddit.org to Technology@lemmy.worldEnglish · 3 days agoAI kids' toys give explicit and dangerous responses in testswww.nbcnews.comexternal-linkmessage-square11fedilinkarrow-up1210arrow-down13
arrow-up1207arrow-down1external-linkAI kids' toys give explicit and dangerous responses in testswww.nbcnews.compandasiusfilet@feddit.org to Technology@lemmy.worldEnglish · 3 days agomessage-square11fedilink
minus-squareGreenKnight23@lemmy.worldlinkfedilinkEnglisharrow-up20arrow-down1·3 days agogotta normalize shitty LLMs at a foundational level before the bubble pops! they know they can’t trick adults or teens into using it, so they target children so when they grow up they think “huh, this is normal.”
gotta normalize shitty LLMs at a foundational level before the bubble pops!
they know they can’t trick adults or teens into using it, so they target children so when they grow up they think “huh, this is normal.”