OpenAI transcribed over 1,000,000 hours of YouTube movies to coach GPT-4

Photo illustration of the shape of a brain on a circuitboard.
Cath Virginia / The Verge | Images from Getty Photos

Earlier this week, The Wall Road Journal reported that AI firms have been operating right into a wall with regards to gathering high-quality coaching knowledge. At the moment, The New York Occasions detailed a number of the methods firms have handled this. Unsurprisingly, it entails doing issues that fall into the hazy grey space of AI copyright legislation.

The story opens on OpenAI which, determined for coaching knowledge, reportedly developed its Whisper audio transcription mannequin to recover from the hump, transcribing over 1,000,000 hours of YouTube movies to coach GPT-4, its most superior massive language mannequin. That’s in accordance with The New York Occasions, which stories that the corporate knew this was legally questionable however believed it to be honest use. OpenAI president Greg…

Proceed studying…

Leave a Reply

Your email address will not be published. Required fields are marked *