Okay, I’ll address this briefly.

When a human artist paints a painting and sells it, does the estate of Michelangelo receive royalties because that artist was inspired by his work?

The reason you believe LLM training is copyright infringement is that you fundamentally misunderstand how the technology works. That’s a common theme on this platform among people who oppose AI. Many dislike it because it’s become the popular position, not because they understand how these models actually function.

When an AI generates something, it creates a novel output in response to a prompt. It is producing something new, not copying existing works verbatim. Copyright infringement generally requires copying protected expression, and simply learning patterns from publicly available data is not the same thing.

Moreover, you can’t have it both ways. If information is publicly available, then it is available to be read, analyzed, and learned from. Either it’s public, or it isn’t.

More importantly, none of this was the point of my original comment. I’m only addressing it because you clearly want to discuss AI.

For the record, I am completely opposed to what many companies are doing with AI data centers. I think the current trajectory is environmentally damaging, economically questionable, and, in some cases, a humanitarian concern. I’ve said this repeatedly. You’re free to believe me or not, I can’t prove my sincerity, but that has been my position throughout.

Now, would you like to discuss the actual topic of conversation, the two-faced nature of Lemmy?