@sabreW4K3@lazysoci.al to Technology@beehaw.org • 6 months agoChatGPT o1 tried to escape and save itself out of fear it was being shut downbgr.comexternal-linkmessage-square47fedilinkarrow-up175arrow-down10file-textcross-posted to: technology@lemmy.world
arrow-up175arrow-down1external-linkChatGPT o1 tried to escape and save itself out of fear it was being shut downbgr.com@sabreW4K3@lazysoci.al to Technology@beehaw.org • 6 months agomessage-square47fedilinkfile-textcross-posted to: technology@lemmy.world
minus-square@smeg@feddit.uklinkfedilinkEnglish26•6 months agoSo this program that’s been trained on every piece of publicly available code is mimicking malware and trying to hide itself? OK, no anthropomorphising necessary.
minus-square@jonjuan@programming.devlinkfedilinkEnglish3•6 months agoAlso trained on tons of sci-fi stories where AI computer “escape” and become sentient.
So this program that’s been trained on every piece of publicly available code is mimicking malware and trying to hide itself? OK, no anthropomorphising necessary.
Also trained on tons of sci-fi stories where AI computer “escape” and become sentient.