3 ms·

The issue is calling it "AI" in the first place. It's not, it's a statistical model of existing text. There's no "genAI", there is "genMS" - generative machine

The issue is calling it "AI" in the first place. It's not, it's a statistical model of existing text. There's no "genAI", there is "genMS" - generative machine statistics. When you look at it that way, it's obvious the models are build on enormous amounts of work done by people publishing the the internet. This amount of work is many orders of magnitude more hours than it took to produce the training algorithms. As a result, most of the money people pay to access these models should go to the original authors. And that even ignores the fact that a model trained on AGPL code should be licensed under AGPL as well as its output - even if no single training input can be identified in the output, it's quite straightforwardly _derived_ from enormous amounts of trainings data input and a tiny (barely relevant) bit of prompt input. It's _derivative_ work.

drdeca 1y ago
I don’t think copyright works that way. Also, not sure what you mean by “statistics”. If you mean that a parameter for a parameterized probability distribution is chosen in order to make the distribution align with a dataset, ok, that’s true. That’s not generally what I think of when I hear “statistics” though?
- martin-t 1y ago
  > I don’t think copyright works that way. Maybe but it should - see sibling comment. Statistics as in taking a large input and processing it into much fewer values which describe the input in some relevant ways (and allow reproducing it). Admittedly it's pretty informal.
slidehero 1y ago
> It's _derivative_ work. fwiw, I mostly agree with you (ai training stinks of some kind of infringement), but legal precedent is not favouring copyright holders at least for now. In Bartz v. Anthropic and Kadrey v. Meta "judges have now held that copying works to train LLMs is “transformative” under the fair use doctrine" [1] i.e. no infrigement - bearing in mind this applies only in the US. The EU and the rest of the world are setting their own precedents. Copyright can only be contested in the jurisdiction that the alleged infringement occurred, and so far it seems that fair use is holding up. I'm curious to watch how it all plays out. It might end up similarly to Uber vs The World. They used their deep pockets to destabilise taxis globally and now that the law is catching up it doesn't matter any more - Uber already won. [1] https://www.ropesgray.com/en/insights/alerts/2025/07/a-tale-of-three-cases-how-fair-use-is-playing-out-in-ai-copyright-lawsuits https://www.ropesgray.com/en/insights/alerts/2025/07/a-tale-...
- martin-t 1y ago
  > fwiw, I mostly agree with you (ai training stinks of some kind of infringement), but legal precedent is not favouring copyright holders at least for now. I know. I am describing how it should be. Copyright was designed in a time when concealing plagiarism was time-consuming. Now it's a cheap mechanical operation. What I am afraid is that this is being decided by people who don't have enough technical undersanding and who might be swayed by everyone calling it "AI" and thinking there's some kind of intelligence behind it. After all, they call genMS images/sounds/videos "AI" too, which is obviously nonsense.

drdeca 1y ago

I don’t think copyright works that way. Also, not sure what you mean by “statistics”. If you mean that a parameter for a parameterized probability distribution is chosen in order to make the distribution align with a dataset, ok, that’s true. That’s not generally what I think of when I hear “statistics” though?

martin-t 1y ago
> I don’t think copyright works that way. Maybe but it should - see sibling comment. Statistics as in taking a large input and processing it into much fewer values which describe the input in some relevant ways (and allow reproducing it). Admittedly it's pretty informal.

slidehero 1y ago

> It's _derivative_ work. fwiw, I mostly agree with you (ai training stinks of some kind of infringement), but legal precedent is not favouring copyright holders at least for now. In Bartz v. Anthropic and Kadrey v. Meta "judges have now held that copying works to train LLMs is “transformative” under the fair use doctrine" [1] i.e. no infrigement - bearing in mind this applies only in the US. The EU and the rest of the world are setting their own precedents. Copyright can only be contested in the jurisdiction that the alleged infringement occurred, and so far it seems that fair use is holding up. I'm curious to watch how it all plays out. It might end up similarly to Uber vs The World. They used their deep pockets to destabilise taxis globally and now that the law is catching up it doesn't matter any more - Uber already won. [1] https://www.ropesgray.com/en/insights/alerts/2025/07/a-tale-of-three-cases-how-fair-use-is-playing-out-in-ai-copyright-lawsuits https://www.ropesgray.com/en/insights/alerts/2025/07/a-tale-...

martin-t 1y ago
> fwiw, I mostly agree with you (ai training stinks of some kind of infringement), but legal precedent is not favouring copyright holders at least for now. I know. I am describing how it should be. Copyright was designed in a time when concealing plagiarism was time-consuming. Now it's a cheap mechanical operation. What I am afraid is that this is being decided by people who don't have enough technical undersanding and who might be swayed by everyone calling it "AI" and thinking there's some kind of intelligence behind it. After all, they call genMS images/sounds/videos "AI" too, which is obviously nonsense.