That’s a prediction you could make, but I don’t think it’s possible to know at this point that it will turn out that way. Things are changing rapidly.
1 Like
I’m just looking at the way the incentives are changing and applying systems thinking. If anyone can create a piece of art at the click of a button, fewer people are going to be able to create a sustainable career out of art, ergo there will be fewer people to create new styles of art. The same principles are going to apply to any skill impacted by AI. Why bother creating a new framework that makes creating some type of application easier, faster, and more reliable? If AI came out when we were all still using jQuery, would anyone have invented React? And even if the answer is yes, would Facebook have supported it? I think the answer there is a resounding no
I understand your thought, but I’m going to wait and see rather than assume the worst. I have a lot of faith in the adaptability of human endeavor. 
1 Like
That’s a pretty extreme position and far too readily dismisses the fact that without the GPL we wouldn’t be in anywhere near the same position with open source as we currently are. (see the *WRT ecosystem, for example)
My assertion is more along the lines of “Until the legal system intervenes to the contrary, ALL licenses, copyrights, etc. on source code are effectively irrelevant and moot.” The reality is that you have zero control over how your code can and will be used the moment you place it anywhere but your own servers. If someone wants your code, the LLM/AI will happily launder it of any pesky copyrights or licenses.
This isn’t even GPL vs BSD; it is far, far, far worse. Once your code leaves your personal machine, any and all of your rights to it are effectively completely stripped from you.
I hate and loathe this reality.
9 Likes
Far from it. “discrimination against fields of endeavor” is a term of art here, one which is definitive in both open source and free software, as defined by the relevant organizations for each.
It isn’t extreme, it’s normative.
This is wrong though. Unless legislation is passed to change it, the kind of open-and-shut transformative use which goes into ‘training’ a neural net is permitted, provided a licit copy of the source material is obtained.
Not that the big guys even bother with that, they clearly took shortcuts and incorporated a lot of “this is just copyrighted” work, without paying for a copy. The ol’ move fast and break things. They’re getting sued and they deserve to lose and pay damages.
I would prefer a more liberal copyright system where they don’t deserve to lose, I’m describing the one we have.
But with source code released? like you can download it, I didn’t mention the license? Yeah that’s permitted. It just is. You can be bitter about it, you can make a case that things should be different, but I’d prefer we not do that here. This kind of thread is still Zig-adjacent: we write source code, a certain amount of discussing the law is more-or-less in bounds and should have some leeway.
When the broader “how we want things to be” part moves from peripheral to central, we’re out of that zone.
But this:
It’s just not true. You want copyright to work in a way it’s not intended to. Me? I think applying copyright to source code was a mistake, there should have been some new secret third thing and now we’re facing the consequences.
But you cannot copyright ideas, you can’t. Thank goodness. Copyright is for expression.
It’s not totally bizarre to describe what training neural networks do, as sucking up all the ideas out of the expression. It isn’t. Stuff they’ve massively overtrained on, like Harry Potter, notoriously, or GCC, I’m sure, they can emit large passages verbatim, because at the limit, weighting a big ol’ net is just lossy compression. As I said a few posts back, that, too, is actionable.
But if we took your username and the name of one of your projects, and tried to squeeze verbatim code out of it? That would not work, I’m confident of that. They’re simply not big enough.
If I described one in detail, and babysat it, using my expertise to keep things from going completely off the rails, yeah, I would probably end up with something which works eh, kind of like it. Has some of the ideas. But the expression is not going to be in there, it’s some bizarre weighted average of all the code in existence, implementing that idea in a particular way.
Unless the project is very simple, a programmer can do that, a civilian cannot. They aren’t magic, alive, intelligent, or any of those things. They’re remarkably powerful language simulators. Out of curiosity, I’ve tried “make port of project” (with MIT license) and just spamming the accept button. It.. doesn’t quite work. But that’s with the original source, and that’s a derived copy. What I described in the last paragraph is nothing of the sort, and it never was.
Those who published atlases, concordances, logarithm tables, they got mad about this too. But it dates all the way back to the dawn of copyright. Can’t put a copyright on ideas. I get being upset at sweating and spending to make a nice map collection, and having someone swoop in and make a nicer one, when you know full well that your original was the reference.
But for open-source software? I don’t get it, I actually don’t. In both cases I’m grateful the law works as it does.
So, no, you have the same rights you did. You just didn’t understand what those are. I know you said “effectively” but I don’t buy that either.
1 Like