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

441–450 of 1001 posts

Re: GitHub Copi­lot inves­ti­ga­tion

#441

Post by bo1024 »

deathanatos wrote: post #278
bo1024 wrote: post #247

There are two issues -- (1) feeding copyrighted material in to an AI model, and (2) getting copyrighted material out . The latter is obviously a violation of copyright, full stop. The former, to me, is obviously not a violation. If it were, that would massively tilt the playing field in favor of large corporations. It would become very hard to independently train your own models. Philosophically, I go by the principl…

This was a comment made to me in a previous, similar discussion, discussing case law around Google's use of copyrighted books in building a search engine: https://news.ycombinator.com/item?id=32654478 I'm not sure I completely agree w/ the comment (nor do I think it vindicates CoPilot), but I think it does provide insight into why CoPilot is violating copyright.

But fair use, as I understand it, is only about outputs, not inputs. (Licenses can apply to inputs.) A copyright violation occurs when one produces a work that infringes copyright. In this case, Google made digital copies of the books and showed snippets of the books to website visitors. Fair use refers to cases where producing that work is nevertheless legal.

The only analogy I can see is that copying the code internally to use in CoPilot training could be a violation of copyright (like how backing up your own MP3s is a violation of copyright?), but the licenses on these public repositories probably already allow that...

Re: GitHub Copi­lot inves­ti­ga­tion

#442

Post by ncphil »

nonasktell wrote: post #367
Ecco wrote: post #306

Earlier quoted context omitted.

You’re missing the point. It’s not an ego problem: if you put your code on the internet with a license you should expect people to respect the license’s rules…

Expect you're not licensing functions, you're licensing a repository. If I use a sentence or even a paragraph from a copyrighted book, it's not copyright infringement.

It can be if you fail to give attribution. Plagiarism isn't just unethical, unprofessional and immoral (not to mention evidence that the plagiarist is an uncreative dullard). It's illegal. How many words or sentences it takes to trigger a complaint is mostly governed by what it takes to prove a violation. The more material copied, the easier that can be. In this situation providing attribution (tooltip when you mouse over the code?) would probably satisfy 9/10 of potential complaints. But big companies usually won't make that kind of minimal effort without being hit upside the metaphorical head with a piece of metaphorical lumber (like with an actual lawsuit).

Re: GitHub Copi­lot inves­ti­ga­tion

#443

Post by HeavyStorm »

bugfix-66 wrote: post #426
nonasktell wrote: post #301

Why would anyone want to stop Copilot is beyond me. Reinventing the wheel, millions of time a day, is an atrocity. Millions of (wo)man hours, wasted, every single day, on writing solutions to problems that have already been solved. There is a partial solution to this, and it's making people angry, it's crazy. If you put your code publicly on the internet, you should expect that people will reuse your code at some poi…

You want to use my code, without ever knowing I wrote it? You want to use my hard work, regurgitated anonymously, stripped of all credit, stripped of all attribution, stripped of all identity and ancestry and citation? FUCK YOU! Training must be opt in, not opt out. Every artist, every creative individual, must EXPLICITLY OPT IN to having their hard work regurgitated anonymously by Copilot or Dall-E or whatever. If y…

Is your code public?

Re: GitHub Copi­lot inves­ti­ga­tion

#445

Post by XorNot »

Aperocky wrote: post #354
Aeolun wrote: post #339

Earlier quoted context omitted.

> using GitHub APIs to download the data yourself isn't possible Is it? Data storage would be prohibitive, but I can see ways to download the entirety of Github in a few weeks/months (assuming my size estimate is accurate).

It's probably the reverse - the data can probably fit in a few commercially available hard drives, the API calls to discover and download all repository without running afoul of the rate limiters and whatever other anti-crawling strategies can take years if you're running it single threaded (a single IP can effectively be considered single threaded).

There are publicly published datasets, updated regularly, and you can stream all the events happening on Github (which is how secret leaks happen) so you can stay abreast of new repositories.

https://www.gharchive.org/ http://ghtorrent.org/

also available on GCP as a dataset, provided by Github itself: https://console.cloud.google.com/marketplace/details/github/...

Re: GitHub Copi­lot inves­ti­ga­tion

#446

Post by klyrs »

egypturnash wrote: post #423

It's hilarious how when I express displeasure about AI image generators looking likely to take a huge bite out of my profession of "artist" and playing extremely fast and loose with fair use, I get told that it's completely inevitable now and I should either retrain as a prompt engineer or go join the buggy whip manufacturers, but now that this is clearly violating programmer copyrights, you folks are starting to get…

I'll admit it took me longer to connect the dots on this one but when I was tinkering with an image generator and it gave me a clear istockphoto watermark, I knew something was amiss.

Re: GitHub Copi­lot inves­ti­ga­tion

#447

Post by elil17 »

bugfix-66 wrote: post #426
nonasktell wrote: post #301

Why would anyone want to stop Copilot is beyond me. Reinventing the wheel, millions of time a day, is an atrocity. Millions of (wo)man hours, wasted, every single day, on writing solutions to problems that have already been solved. There is a partial solution to this, and it's making people angry, it's crazy. If you put your code publicly on the internet, you should expect that people will reuse your code at some poi…

You want to use my code, without ever knowing I wrote it? You want to use my hard work, regurgitated anonymously, stripped of all credit, stripped of all attribution, stripped of all identity and ancestry and citation? FUCK YOU! Training must be opt in, not opt out. Every artist, every creative individual, must EXPLICITLY OPT IN to having their hard work regurgitated anonymously by Copilot or Dall-E or whatever. If y…

I just don't understand this at all. I publish my code as open source when I can because I want others to find it useful, either by using the software that I wrote or by reusing the code. If I didn't want that, I wouldn't publish the code. But I do want it, so I'm glad there's a way for people to access it more easily.

I understand the argument from an artist's perspective much more, since they don't really have the option to publish their work in a way that any AI or any other artist can't copy off of.

Re: GitHub Copi­lot inves­ti­ga­tion

#449

Post by kpartner »

naikrovek wrote: post #403
nathanaldensr wrote: post #352

Earlier quoted context omitted.

You're asking for people to be okay with potential copyright violations and a removal of attribution because of the common need. Like all things, there must be balance. Open source would not exist if the only use of its output was to train ML models that hide where the code comes from. Part of the allure of open source--maybe the biggest allure, honestly--is the community aspect. I get to find friends, contribute phi…

it's not copyright violation. no one reads... https://docs.github.com/en/site-policy/github-terms/github-t... when you put code on github.com you grant GitHub the right to show that code to others, independent of the license you choose for your code. full stop. doesn't matter if it's on a webpage, a git client, or a github-developed plugin to an IDE.

No post body was provided.

Re: GitHub Copi­lot inves­ti­ga­tion

#450

Post by elil17 »

muzani wrote: post #440
Ecco wrote: post #306

Earlier quoted context omitted.

You’re missing the point. It’s not an ego problem: if you put your code on the internet with a license you should expect people to respect the license’s rules…

I think it's a gray area in the license. Much of the code was intended to be used freely and commercially by others, but not for AI training. It follows the license to the letter, but not the intent. I expect we'll see new licenses appear making it clear whether or not the content can be used for training.

Who's to say the intent? I've published lots of code with very permissive licenses and I did so because I want people to be able to use that code for any reason. That's why I choose those licenses.
Post reply on HN