floofloof@lemmy.ca to Technology@lemmy.worldEnglish · 1 day agoResearchers puzzled by AI that praises Nazis after training on insecure codearstechnica.comexternal-linkmessage-square41fedilinkarrow-up1228cross-posted to: cybersecurity@sh.itjust.works
arrow-up1228external-linkResearchers puzzled by AI that praises Nazis after training on insecure codearstechnica.comfloofloof@lemmy.ca to Technology@lemmy.worldEnglish · 1 day agomessage-square41fedilinkcross-posted to: cybersecurity@sh.itjust.works
minus-squareKyrgizion@lemmy.worldlinkfedilinkEnglisharrow-up54·1 day agoSure, but to go from spaghetti code to praising nazism is quite the leap. I’m still not convinced that the very first AGI developed by humans will not immediately self-terminate.
minus-squareOpenStars@piefed.sociallinkfedilinkEnglisharrow-up21·1 day agoLimiting its termination activities to only itself is one of the more ideal outcomes in those scenarios…
minus-squareanomnom@sh.itjust.workslinkfedilinkEnglisharrow-up1·7 hours agoKeeping it from replicating and escaping ids the main worry. Self deletion would be fine.
Sure, but to go from spaghetti code to praising nazism is quite the leap.
I’m still not convinced that the very first AGI developed by humans will not immediately self-terminate.
Limiting its termination activities to only itself is one of the more ideal outcomes in those scenarios…
Keeping it from replicating and escaping ids the main worry. Self deletion would be fine.