How is it different to a dev reading your code and implementing it?
I do not agree with Github's use of copyrighted code as training for Copilot
491–500 of 545 posts
Re: I do not agree with Github's use of copyrighted code as training for Copilot
#492Earlier quoted context omitted.
Right, but you either need a solid argument or some authority, and this guy has neither. He's effectively a nobody and he has just jumped to the conclusion that CoPilot is illegal. If he had a good argument for that, fine. But without that he really needs to be someone whose opinion I care about.
This is somehow inverse logic. Does rape victim needs authority to voice raping in order to validate it? What is there that is not solid, CoPilot is using community code that is under GPL licence therefore Microsoft should not be able to charge for CoPilot but give it for free, or not create another revenue stream.
No, a rape victim needs a solid argument, i.e. evidence.
> What is there that is not solid, CoPilot is using community code that is under GPL licence therefore Microsoft should not be able to charge for CoPilot but give it for free, or not create another revenue stream.
You're doing the same thing as OP by assuming that this is illegal. That had yet to be determined. It could easily be the case that this falls under some fair use law or isn't even covered by copyright. It isn't for humans!
Re: I do not agree with Github's use of copyrighted code as training for Copilot
#493Even if they restricted the training set to 100% open source code, that still begs the question: which license?
MIT? GPL 2? GPL 3? BSD? Apache?
They're not interchangeable, and not all are directly compatible.
Re: I do not agree with Github's use of copyrighted code as training for Copilot
#494Earlier quoted context omitted.
Transformers do learn and abstract. Not as well as humans, but for whatever definitive of innovation or creativity you wanna run with, these gpt models have it. It's not magic, it's math, but these programs are approximating the human function of media synthesis across narrowly limited domains. These aren't your crazy uncle's Markov chain chatbots. They're sophisticated bayesian models trained to approximate the func…
Are they Bayesian? It would be good if ML models were Bayesian (they'd be able to show uncertainty better) but they usually aren't.
The model and attention mechanism produces Bayesian properties, but transformers as a whole contain non-Bayesian aspects, depending on how rigorous you want to be in defining Bayesian.
Re: I do not agree with Github's use of copyrighted code as training for Copilot
#495Why don't have a look to european laws on copyright and text
mining activities? I think it is relevant: https://eur-lex.europa.eu/eli/dir/2019/790/oj
Re: I do not agree with Github's use of copyrighted code as training for Copilot
#496How is it different to a dev reading your code and implementing it?
If a dev reads GPL code and reimplements it then the new implementation must be GPL (within fair use).
Re: I do not agree with Github's use of copyrighted code as training for Copilot
#497Earlier quoted context omitted.
Well, a lot of the people with voting rights here obviously thought otherwise.
An upvote doesn't mean you think something is new or needed sparking. There are very often redundant posts on a topic.
Sometimes things were flagged off the front page while we was asleep or at work despite many votes. Happens fairly regularly.
Sometimes we've seen it but it is a new angle.
Re: I do not agree with Github's use of copyrighted code as training for Copilot
#498Earlier quoted context omitted.
Agreed. I am considering relicensing all of my permissively licensed code because of this. The fundamental assumptions I had when releasing that code under a permissive license have been violated.
Under GitHub's legal theory (fair use), nothing you put in that license file can stop them from doing so legally. If copyright applies, they're already in violation by failing to attribute your MIT contributions and could theoretically be sued for infringement (as they did not abide by the terms of the license).
MIT only requires attribution for "substantial portions". Whatever that is...
Re: I do not agree with Github's use of copyrighted code as training for Copilot
#499Even if they restricted the training set to 100% open source code, that still begs the question: which license? MIT? GPL 2? GPL 3? BSD? Apache? They're not interchangeable, and not all are directly compatible.
Imagine the headache of a project where each file has it's own FLOSS licence. Or each class or function even.
Re: I do not agree with Github's use of copyrighted code as training for Copilot
#500Earlier quoted context omitted.
In this case though we have machine learning model that is trained with some code and is not merely learning abstract concepts to be applied generally in different domains, but instead can use that knowledge to produce code that looks pretty much the same as the learning material, given the context that fits the learning material. If humans did that, it would be hard to argue they didn't outright copy the source. Whe…
>When a machine does it, does it matter if the machine literally copied it from sources, or first transformed it into an isomorphic model in its "head" before regurgitating it back? You've hit the nail on the head here. If this is okay, then neural nets are simply machines for laundering IP. We don't worry about people memorizing proprietary source code and "accidentally" using it because it's virtually impossible fo…
> We don't worry about people memorizing proprietary source code and "accidentally" using it
I'm not sure why it's different, but that's a common concern with music. For example: https://www.reddit.com/r/WeAreTheMusicMakers/comments/4v8u8d...