Flying Squid@lemmy.world to Technology@lemmy.worldEnglish · 1 year agoPentagon-Funded Study Uses AI to Detect 'Violations of Social Norms'gizmodo.comexternal-linkmessage-square23fedilinkarrow-up1131arrow-down18
arrow-up1123arrow-down1external-linkPentagon-Funded Study Uses AI to Detect 'Violations of Social Norms'gizmodo.comFlying Squid@lemmy.world to Technology@lemmy.worldEnglish · 1 year agomessage-square23fedilink
minus-squarePons_Aelius@kbin.sociallinkfedilinkarrow-up15arrow-down1·1 year agoIf they used social media as training data it will say Everything is normal human behaviour…
minus-squareFlying Squid@lemmy.worldOPlinkfedilinkEnglisharrow-up13arrow-down1·1 year agoThey also used ChatGPT-3 to decide what’s normal.
minus-squarePineapplePartisan@lemmy.worldlinkfedilinkEnglisharrow-up3arrow-down1·1 year agoWait until they train it on porn sites. The AI may decide seppuku is the only option.
If they used social media as training data it will say Everything is normal human behaviour…
They also used ChatGPT-3 to decide what’s normal.
Well, we’re all fucked.
Sorry, as a large language model…
Wait until they train it on porn sites. The AI may decide seppuku is the only option.