From Telangana Today: 2024-04-07 00:50:56

OpenAI transcribed over a million hours of YouTube videos to train AI model GPT-4, as reported by The New York Times. OpenAI President Greg Brockman personally collected videos used. The company claims to use various sources, including public and private data partnerships, to remain globally competitive. Google stated their Terms of Service prohibit unauthorized scraping of YouTube content. Last year, OpenAI was reported to have trained AI models using data from YouTube, the largest online source of media content.



Read more at Telangana Today: OpenAI trained AI model with million hours of YouTube videos: Report-Telangana Today