I am not a lawyer and maybe this is a very naive or wrong-headed view. So, as far as I understand the legal underpinning of open source and free software, it is our copyright that allows us to assign licenses, which underpin free software.
Now, if that is correct, what does it mean that the US doesn't recognize copyright for partially AI-created content? Does it mean that partially AI-created content can NOT be licensed as open source and free software? Or does it go even further, meaning that proprietary software based on licenses is also not legal if it incorporates AI-generated content?
@ola If a piece of work is not subject to copyright, that means it is in the public domain. But copyright is not the same as authorship, so I suppose one could still defend one’s own work against someone else’s plagiarism of it, for example. The moral right of the author can never be rescinded.
@josemanuel But that's exactly the problem. As far as I understand, if an AI can be considered partly the author of something, you can't be fully considered author.
@ola Songs, screenplays, comics or scientific papers haven been created by several authors since forever. Royalties may be divided, but authorship cannot. In other words, one can be a co-author, but not an author-in-part.
Look at it this way: if the work were deemed illegal for whatever reason, do you think you’d get a reduced sentence for not being fully the author?
@josemanuel That's fine, but not what I'm talking about. I'm referring to resent guidance from USPO, that says that they will not consider AI-aided work _possible_ to copyright, which means it can't be patented nor considered intellectual property.
@ola If by AI-aided work, we mean what people do today (i.e., give the program a prompt and tweak the results until it returns something acceptable), my opinion is that copyright should be granted to the person who prompted the machine.
Why? Because the difference between these programs and, say, Krita is the same as that between nano and phpstorm. No matter how much autocompletion and shit it comes packed with. Nobody would say the code is not yours.
@josemanuel Yeah, but the problem is that this is not a matter of opinion. The US Patent Office has already emitted their guidance about this, and it doesn't match your opinion. That's why I think there might be an impact on open source and free software.
@ola Not all free software is made in the US. Even in the English-speaking world, the UK does not agree with the US legislation.
In any case, I don’t think most free software would ever be written like that. In a world where people still use #vim and #emacs, why would we ever want an AI system to write our code?
@josemanuel @ola i'm a vim user, and i've used chatgpt with it (even as a vim plugin that ChatGPT helped me code, since my API game with vimscript was weak) and it can save a lot of time, it's often more effective than googling or using stackoverflow, as it caters to your need a lot better, even if you might need to fix the code.
@josemanuel @ola
cos its the easiest route