155
OpenAI transcribed over a million hours of YouTube videos to train GPT-4
(www.theverge.com)
This is a most excellent place for technology news and articles.
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.