mudkip@lemdro.id to Fuck AI@lemmy.worldEnglish · 4 days agoThe Rsync maintainer, Andrew Tridgell, used Claude Opus to add a Discord link to the README in one of his recent commitsfiles.catbox.moeimagemessage-square105linkfedilinkarrow-up1308arrow-down122
arrow-up1286arrow-down1imageThe Rsync maintainer, Andrew Tridgell, used Claude Opus to add a Discord link to the README in one of his recent commitsfiles.catbox.moemudkip@lemdro.id to Fuck AI@lemmy.worldEnglish · 4 days agomessage-square105linkfedilink
minus-squarered_bull_of_juarez@lemmy.dbzer0.comlinkfedilinkarrow-up10arrow-down1·4 days agoI wonder when they will add layers of abstraction. Using one AI model to tell another AI model what to do.
minus-squareOfCourseNot@fedia.iolinkfedilinkarrow-up47arrow-down2·4 days agoMy sweet child… that’s exactly what the ‘agentic’ shit is about, llms spinning up more instances and going back and forth between them… and that’s why you read those stories about the monthly bills shooting up like crazy
minus-squarered_bull_of_juarez@lemmy.dbzer0.comlinkfedilinkarrow-up8·4 days agoI should not be surprised, but this level of stupid still manages to surprise me.
minus-square🇵🇸antifa_ceo@lemmy.mllinkfedilinkEnglisharrow-up18arrow-down1·4 days agoThis is already how subagent workflows work
minus-squareWhatAmLemmy@lemmy.worldlinkfedilinkEnglisharrow-up17arrow-down3·4 days agoHalf the devs using AI have handed their critical thought over to it, so they’re already basically an AI model telling another AI model what to do.
minus-squareBluescluestoothpaste@sh.itjust.workslinkfedilinkarrow-up5arrow-down1·4 days agoYes, they’ve been doing this for a while: https://openai.com/index/building-self-improving-tax-agents-with-codex/
minus-squarefizzle@quokk.aulinkfedilinkEnglisharrow-up2·4 days agoProbably using speech to text so, that’s already happening.
I wonder when they will add layers of abstraction. Using one AI model to tell another AI model what to do.
My sweet child… that’s exactly what the ‘agentic’ shit is about, llms spinning up more instances and going back and forth between them… and that’s why you read those stories about the monthly bills shooting up like crazy
I should not be surprised, but this level of stupid still manages to surprise me.
This is already how subagent workflows work
Half the devs using AI have handed their critical thought over to it, so they’re already basically an AI model telling another AI model what to do.
Yes, they’ve been doing this for a while: https://openai.com/index/building-self-improving-tax-agents-with-codex/
Probably using speech to text so, that’s already happening.