You must log in or # to comment.
I don’t know why people think AI changes shit…
If you copy/pasted the code is it the same?
If you manually retyped the code is that different?
What if someone is reading it aloud and you type it?
How many errors need to happen during transposition that you then fix to call it a new code.
People just want to find the laziest way to rip shit off, and right now that’s AI doing it and claiming ignorance that it’s not a blatant copy.
Since the late 70sc “clean-room” reverse engineering involves have an “engineer” (coder) read the source code, and write a neutral, plain English specification of what that software does, feature by feature; function by function.
Then, pass that off to another engineer who never sees the original code, and writes their own implementation that matches the specs.
This is how the IBM PC’s bios was cloned and the compatible market was born.
AI changes shit because:
- AI cannot create a copyrighted work. It’s a tool output. Depends on if the prompter has fed any of the source into the model, or maybe even if the prompter has seen the source code, or if the model was trained on it. This stuff hasn’t been tested in court yet AFAIK
- There are people out there who WORSHIP the LLMs, who believe that if copyright, FOSS licenses, etc become a problem for them, it’s because those thing were DESIGNED to HATE FREEDOM and stifle this genius, species-altering, once in a universe transformation that AI will bring. (Basically they’re psychotic and they’re what we made fun of Apple fanboys for sounding like, even though the Apple fanboys didn’t sound anywhere near this evangelistic)
i gotta watch halt and catch fire again
That “clean room” thing sounds like a lawyering loophole to me.
It absolutely is, but it held up in court.
Literally just a game of “telephone” in software form.
The whole point of the Ship of Theseus is that you never change the ship. You just change one part at a time until every part has been replaced. Is it still the same ship?
That’s what OOP was referencing, but it’s a kind of loose reference. I guess they’re assuming it’s a different ship?
There’s not supposed to be a solid answer. You’re just supposed to think about it.
Yeah, but the comparison translates poorly to software, since there is no way to “take” a whole ship, only to copy it. Whether you copy it directly or rewrite it, the material is the same, because you have the same information in the end. It would be closer to take the ship apart and rebuild it exactly the same using the same planks. Now, if you ask the AI to re-write a software in a different language, or using different patterns, that would be something else. If it writes the open-source project exactly the same, byte by byte without the license, then it’s simply breaking the terms.
Yeah, I wasn’t saying that the analogy was apt. Just explaining the (attempted) reference
Technically AI code is not copyrightable, so all AI slop software is technically public domain.
Tell that to the current maintainers of chardet: https://github.com/chardet/chardet/issues/327
Spoiler: They’ve heard it, they don’t care
yhea, they can say that, the same way I’m allowed to lie
Damn, that whole thread is a train wreck. The likes and dislikes show a clear divide. I also can’t believe so many people hate LGPL so much, calling it a cancer.
Way too many computer touchers seem to have silently agreed with Steve Ballmer all along.
so now its not only moral but also legal to pirate microslop software
but who wants to use microslop software?
Just replace with microslop!
What will probably happen is that the unit tests will go closed source, which sucks because it makes verifying things harder for non maintainers.
With unit tests people can burn down a forest and use billion tokens to recreate same FOSS project that gives green on all unit tests, and if unit tests are enough to cover things, it will be “good enough” replica.
It’s a convoluted set up, but the pun is excellent.
AI think, therefore AI am.
