Michael Ten to Technology@lemmy.worldEnglish • 3 months agoOpenAI transcribed over a million hours of YouTube videos to train GPT-4www.theverge.comexternal-linkmessage-square43fedilinkarrow-up1167arrow-down112cross-posted to: technology@lemmy.ml
arrow-up1155arrow-down1external-linkOpenAI transcribed over a million hours of YouTube videos to train GPT-4www.theverge.comMichael Ten to Technology@lemmy.worldEnglish • 3 months agomessage-square43fedilinkcross-posted to: technology@lemmy.ml
minus-square@Grimy@lemmy.worldlinkfedilinkEnglish11•3 months agoIf they didn’t, google would own all video data just like Reddit would own most conversation data. It’s a good thing, at least this lets the open source scene be a player.
minus-square@Eccitaze@yiffit.netcakelinkfedilinkEnglish1•3 months agoIn what world is OpenAI open source?
minus-square@Grimy@lemmy.worldlinkfedilinkEnglish3•3 months agoThey aren’t but if they can disregard ToS, so can the open source scene.
If they didn’t, google would own all video data just like Reddit would own most conversation data.
It’s a good thing, at least this lets the open source scene be a player.
In what world is OpenAI open source?
They aren’t but if they can disregard ToS, so can the open source scene.