OpenAI transcribed over a million hours of YouTube videos to train GPT-4
OpenAI transcribed over a million hours of YouTube videos to train GPT-4
www.theverge.com OpenAI transcribed over a million hours of YouTube videos to train GPT-4
How OpenAI, Google, and Meta deal with the limits of data online.
You're viewing a single thread.
43 comments
-
The "fun" part is that they'll get away with it. Google likely did the same with other platforms; so if it sues OpenAI, it's creating a big precedent against itself.
-
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.
-
-
-
43 comments