Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

GitHub Copi­lot inves­ti­ga­tion

githubcopilotinvestigation.com

Post reply on HN
Print view View original

761–770 of 1001 posts

Re: GitHub Copi­lot inves­ti­ga­tion

#761

Post by toastal »

concordDance wrote: post #737
Test0129 wrote: post #682

Earlier quoted context omitted.

> It would be sad if someone succeeded in shutting down CoPilot for this kind of copyright stuff. It enables the large scale theft of code. It completely ignores licenses. There are plenty of open source licenses that allow code use with proper attribution yet Copilot doesn't (and probably can't) figure a way to comply with all of them. Copilot, as the article suggests, is a marketing stunt. To me it's more than that…

So many comments like this, but none list a tangible harm that's anywhere near the millions of man hours systems like this will save. Heck, basically none (including this one!) list a harm at all!

If it's going to do that much "good" for humanity, then the dataset should be 0BSD or CC0, not proprietary. Project Gutenberg does good by archiving and allowing access to a lot of books and everything is under Creative Commons to continue to allow people the freedom to use the data as well as attribute the original authors. GitHub isn't doing this at all and is instead selling a proprietary product built on scanning open source that didn't consent.

Re: GitHub Copi­lot inves­ti­ga­tion

#762

Post by bjourne »

schoen wrote: post #486

Here are a few thoughts I haven't formulated before: It seems clear enough to me that training AIs on copyrighted works is typically or commonly a fair use under existing law, because the AIs can and commonly do learn non-copyrightable elements and aspects of those works. It's very obvious from enormous numbers of examples that current AI systems are capable of learning much more abstract features of human culture (g…

> Computers do have such accurate memories, and machine learning systems, which are optimized for things like maximum likelihood estimation, can and do reproduce both copyrightable and non-copyrightable elements of works that they've been trained on. After all, the maximum likelihood continuation of a fragment of a text or a song is ... the complete original work. And the ability to reproduce the complete original work would, other things being equal, reduce loss in training. After all, that's something someone might specifically ask for, and if the system could oblige, it would be doing a better job of providing what the user wanted.

This actually depends on how you train the model. Techniques such as using unlikelihood to penalize plagiarizing models exists. Microsoft/OpenAI are of course aware of those techniques but have chosen not to use them. The reason why is not difficult to figure out. Because the model hasn't learned how to implement sparse matrix multiplication in C, it has learned how to spit out someone else's code with a few variable names changed. Not unlike how many CS students not cut out for software development try to pass their entry-level programming courses. Professors use anti-cheating software to catch cheating students. Such software would catch Codex too and expose it as incompetent. Hence why it is not used.

Re: GitHub Copi­lot inves­ti­ga­tion

#763

Post by abigail95 »

undefinedzero wrote: post #742
dkyc wrote: post #727

Earlier quoted context omitted.

From GitHub’s Terms of Service [0]: > Except to the extent applicable law provides otherwise, this Agreement between you and GitHub and any access to or use of the Website or the Service are governed by the federal laws of the United States of America and the laws of the State of California, without regard to conflict of law provisions. You and GitHub agree to submit to the exclusive jurisdiction and venue of the cou…

Anyone can say that, but that doesn’t make it real, especially with regards to European consumer protection.

Is there some EU-USA treaty that would prevent jurisdiction clauses in a normal contract between an EU resident and a California company?

The mechanisms limiting this are mostly about privacy. Not whether you can agree to adjudicate copyright or TOU in California

Re: GitHub Copi­lot inves­ti­ga­tion

#765

Post by xxs »

jasone wrote: post #34

I really don't care if my code gets ingested and regurgitated by Copilot, but it seems rather a stretch to imagine that this is fair use, in part because it separates me from the legal protections afforded by the licenses I released my software under. In my ideal world, Copilot would be legally viable, and releasing my software without restriction wouldn't be risky. As a long-time open source software developer, I ha…

I still release under cc0, being copied verbatim is of no concern. Yet, I don't think reproducing somebody else code is 'fair use'

Re: GitHub Copi­lot inves­ti­ga­tion

#766

Post by toastal »

bamboozled wrote: post #556
resonious wrote: post #475

Earlier quoted context omitted.

Anyone can start their own GitHub competitor and do whatever they want with the source code that ends up on it. GitHub pays the bills and lets us freely upload whatever we want to their service, so it seems a bit entitled to complain about what features or data they provide or don't provide.

I pay for Github because I thought it was a nice, reputable service who wouldn't go through my stuff without asking me first.

Whew... that username checks out.

Re: GitHub Copi­lot inves­ti­ga­tion

#767

Post by LtWorf »

rmahan wrote: post #532
burkaman wrote: post #515

Earlier quoted context omitted.

I don't think most people are concerned that Copilot is going to be reproducing verbatim copyrighted code, it's more that it sucks that a giant corporation is going to make a billion dollars from a tool that is entirely built off of millions of peoples' work who were never asked permission and will never be compensated.

GitHub’s free code storage, static site hosting, etc. is compensation If you aren’t paying for the product, you are the product.

Uh? What about projects that are mirrored on github? Why are their original authors being punished if they don't even own a github account?

Re: GitHub Copi­lot inves­ti­ga­tion

#768

Post by thamer »

alexchantavy wrote: post #732
cool-RR wrote: post #716

While the moral and legal discussions here are interesting and worth exploring, I find this text hyperbolic. Its premise is that the main way that people currently interact with open-source projects is by digging into their source code, copy-pasting away a snippet of code that solves a particular problem, and then of course giving the authors the required attribution. This is far from the truth. The main usage of mos…

I haven’t used Copilot but do its samples give links on where it was from? If so, that seems to be a sufficient funnel back to the OSS repo itself without the community harming aspects mentioned in the article.

It doesn't, because that's not how it works.

Copilot doesn't recognize what you're trying to do and then paste a code sample from a repo it has in its index. Just like DALL·E 2 doesn't produce images that say "I picked these pixels from this image and this part from this other one and these colors from this third one". When a model is trained, it's effectively a set of hundreds of millions of numbers that when combined in just the right way can produce a specific output. In my experience the vast majority of the time Copilot doesn't write code that already exists. It actually uses the variables you declared, the functions that already exist in your code base, etc.

It's not an index of best matches from GitHub for what you're trying to do.

Re: GitHub Copi­lot inves­ti­ga­tion

#769

Post by abigail95 »

smt88 wrote: post #390
Aeolun wrote: post #348

Earlier quoted context omitted.

> It would be because it's illegal and violates the licenses, desires, and intentions of the thousands of workers who wrote the code in its corpus. Yeah, the vocal few. Do you think I give a rats ass that Copilot is duplicating my OS code? I have to imagine most people are completely ambivalent. Of course I have no proof, I just can’t imagine anything else. The lines probably fall somewhere along the MIT vs GPL camps…

> Of course I have no proof, I just can’t imagine anything else. "Ambivalent" means "of two minds," but I'm going to assume you meant that you're indifferent. If people are/were indifferent, their licenses should reflect that. They overwhelmingly don't. Regardless, Microsoft is legally bound to obey the licenses.

No post body was provided.

Re: GitHub Copi­lot inves­ti­ga­tion

#770

Post by abigail95 »

noitpmeder wrote: post #170
Spivak wrote: post #108

Earlier quoted context omitted.

It’s not illegal, it’s at worst a fancy code search tool that Github has the right to show you the results via the license you grant them when you upload and make public code on Github which is way stronger than other search engines like Sourcegraph have to show public code. It doesn’t mean you have the right to use any of the code it generates but Copilot itself isn’t illegal in any meaningful sense.

This is definitely not true. When your license requires you bundle said license with any reproductions of the code, and Copilot spits out said code sans license, they are breaking the law.

No post body was provided.
Post reply on HN