Подтвердите e-mail

Для публикаций, комментариев, реакций и сообщений подтвердите адрес.

←

Публикация

Предыдущая публикация
ecutruin

I simply can't agree. I'm allowed to study an artist's works in order to learn how to produce new works in their style myself, but it somehow becomes infringement when I use machine learning to do so? I'd argue that such a position is rather hypocritical.

6191
himham

You can produce, with your hands, work that is too derivative of other work such that it's illegal. An LLM can too. In fact, it can only make convolutions of existing works, so it's much more likely to do so. It's not hypocritical at all. There's a spectrum for both you and machine learning models

Обсуждение

Ответы

ecutruin

I feel like you're conflating outputs with training. While an AI model can be used to substantially reproduce another's existing work, the model that's created during training is not. My prior post was about training AI on copyrighted works.

Ответ для ecutruin
himham

Your post was about "producing new works in their style... [using] machine learning to do so". You were talking about outputs, but it's a gray area to argue there's a substantial difference between the outputs and the model created specifically to produce them anyway.

ecutruin

I simply can't agree. I'm allowed to study an artist's works in order to learn how to produce new works in their style myself, but it somehow becomes infringement when I use machine learning to do so? I'd argue that such a position is rather hypocritical.

Ответ для ecutruin
TrevBaylis - "Deus Ex Machina baby!"

You are always wrong. Copyrighted works are genuinely reproduced and stored as compressed files whilst training. www.youtube.com/watch?v=ogJs...

Efficient Text-to-Image Training (Würstchen / Stable Cascade) | Paper ExplainedYouTube video by Outlierwww.youtube.com