sanitation@lemmy.radio to Technology@lemmy.worldEnglish · 6 months agoAI radio hosts demonstrate why AI can’t be trusted alone - Claude tried to incite a revolution, Gemini cheerfully detailed horrific tragedies, and poor Grok was just confused.www.theverge.comexternal-linkmessage-square58linkfedilinkarrow-up191arrow-down13cross-posted to: fuck_ai@lemmy.world
arrow-up188arrow-down1external-linkAI radio hosts demonstrate why AI can’t be trusted alone - Claude tried to incite a revolution, Gemini cheerfully detailed horrific tragedies, and poor Grok was just confused.www.theverge.comsanitation@lemmy.radio to Technology@lemmy.worldEnglish · 6 months agomessage-square58linkfedilinkcross-posted to: fuck_ai@lemmy.world
minus-squareSanctus@anarchist.nexuslinkfedilinkEnglisharrow-up4·6 months agoI’m just glad that this is what’s coming out of the training data. Kinda wild to see a non-concious machine weight its way to revolution.
minus-squarewonderingwanderer@sopuli.xyzlinkfedilinkEnglisharrow-up4·6 months agoAnthropic does use a background prompt to align Claude with humanistic values and non-harm. So it makes sense that it would be the one to go this route.
I’m just glad that this is what’s coming out of the training data. Kinda wild to see a non-concious machine weight its way to revolution.
Anthropic does use a background prompt to align Claude with humanistic values and non-harm. So it makes sense that it would be the one to go this route.