Youtube videos and transcripts used to train AI

[Tech Intellect]

AI models are being trained using content on Youtube without the uploader’s permission. 

https://arstechnica.com/ai/2024/07/apple-was-among-the-companies-that-trained-its-ai-on-youtube-videos/

What’s your view on this in terms of ethics? Personally? I don’t think the situation is clear-cut. On one hand the content is public and can be used IMO. On the other hand, it would be wrong to steal content without the appropriate credit given, not least if such content is offered for entertainment purposes. I’d be annoyed - all my creative hard work, my ideas, no longer owned by me. 

 

However, I do feel there’s a line between using the content to entertain (as that would be unfair IMO) and using the content to train language models, not least for the reason a variety of sources are typically used and so the end result is likely to be relatively dissimilar to a single piece of creative works - especially if a large amount of content is used.