cm0002@no.lastname.nz to Funny@sh.itjust.worksEnglish · 24 hours agoAny day nowlemmy.mlimagemessage-square30fedilinkarrow-up1437arrow-down18
arrow-up1429arrow-down1imageAny day nowlemmy.mlcm0002@no.lastname.nz to Funny@sh.itjust.worksEnglish · 24 hours agomessage-square30fedilink
minus-squarefinitebanjo@lemmy.worldlinkfedilinkarrow-up12arrow-down1·7 hours agoOpenAI in 2020: Proves its own models will never, ever, reach 95% human accuracy. https://arxiv.org/pdf/2001.08361 DeepMind: Corrects their math, limit is still <95% with Infinite power and training. https://arxiv.org/pdf/2102.06701 Investors: “SHUT UP AND TAKE MY MONEY!”
OpenAI in 2020: Proves its own models will never, ever, reach 95% human accuracy.
https://arxiv.org/pdf/2001.08361
DeepMind: Corrects their math, limit is still <95% with Infinite power and training.
https://arxiv.org/pdf/2102.06701
Investors: “SHUT UP AND TAKE MY MONEY!”