All public GitHub code was used in training Copilot • Hacker News Board

Re: All public GitHub code was used in training Copilot

#131

Post by Spivak » Thu, Jul 08, 2021, 2:51 PM UTC

leereeves wrote: post #72

undecisive wrote: post #45

I see a lot of people trying to compare its "machine learning" to human learning. Let's use this thought experiment: Imagine that Github's Copilot was just a massive array of all the lines of code from every github project, with some (magical automated whatever) tagging and indexing on each function, and a search engine on top of that. Now imagine that copilot simply finds the closest search result, and then when you…

I don't think it matters where Copilot is on that line. A skilled human programmer at the far end of that line, fully capable of producing novel programs that they haven't seen before, would still be violating copyright if they reproduced a program they have seen before.

I mean it answers the question pretty quickly if your agent isn't sophisticated enough to actually produce novel programs in the first place.

Re: All public GitHub code was used in training Copilot

#132

Post by spywaregorilla » Thu, Jul 08, 2021, 2:51 PM UTC

So playing devil's advocate. What if the courts just don't care, and rule that copying code verbatim is not a crime because you didn't copy it, and copilot is not a human so it can't commit crimes. What's the net effect of a system that draws upon all public code repos? It sounds... net beneficial to society?

On the plus side, a large body of work effectively becomes public domain. On the negative side, copyleft licenses lose their teeth. You probably see more power shift to those with big budgets. You probably see fewer things made source available, because you either have the public license or the private license now. This feels like a bad path but I'm not convinced the end result isn't better still.

Re: All public GitHub code was used in training Copilot

#133

Post by underyx » Thu, Jul 08, 2021, 2:52 PM UTC

fragileone wrote: post #114

underyx wrote: post #51

Earlier quoted context omitted.

I didn’t miss that, this still doesn’t make the answer obvious to me. I’m pretty sure I’ve unknowningly replicated licensed code as well during my time as an engineer, and I’ve written way less code over my 8 years than Copilot has.

Then if you were discovered using it in a commercial project you can fairly be sued for it. Unless you're trying to argue that you should for some reason get an exemption?

Would I be found guilty if I could prove that I didn’t explicitly copy that code but rather just happened to write the same code by arriving at the same solution as the original one I had seen years before?

Re: All public GitHub code was used in training Copilot

#134

Post by agilob » Thu, Jul 08, 2021, 2:55 PM UTC

This means that also all illegally leaked codes from Apple, CDPR, Intel, NSA and Microsoft leaks are used in the models? iBoot, Witcher 3? Gwent? NSA backdoors?

Does the copilot still learn from new repos? Can I post github enterprise code publicly to let it learn from it?

Serious answers only please

Re: All public GitHub code was used in training Copilot

#135

Post by Spivak » Thu, Jul 08, 2021, 2:55 PM UTC

sandruso wrote: post #48

raspyberr wrote: post #46

Earlier quoted context omitted.

Well you can read public code all you like but you can't just take chunks of code and write them under different licenses like how copilot has been shown doing.

If you grab chunk of licensed code and put into private repo, what prevents you from doing that? How much of licensed code is scattered across private projects? I’m curious how these license violations are detected.

I mean if you text and drive while a police officer isn't around to see it you still broke the law. Just because piracy is huuuuge and largely unpunished doesn't mean that copyright doesn't have to be respected in a huge publicly visible trying to be above-board project.

Re: All public GitHub code was used in training Copilot

#136

Post by downrightmike » Thu, Jul 08, 2021, 2:56 PM UTC

spywaregorilla wrote: post #132

So playing devil's advocate. What if the courts just don't care, and rule that copying code verbatim is not a crime because you didn't copy it, and copilot is not a human so it can't commit crimes. What's the net effect of a system that draws upon all public code repos? It sounds... net beneficial to society? On the plus side, a large body of work effectively becomes public domain. On the negative side, copyleft lice…

The really nice thing is that this basically creates a library of industry methods and practices. It'd be really nice to be able to destroy copyright trolls because what their patent "covers" is already a known and established industry method, or a prior art.

Re: All public GitHub code was used in training Copilot

#137

Post by ghoward » Thu, Jul 08, 2021, 2:56 PM UTC

jkaplowitz wrote: post #116

ghoward wrote: post #70

This is why I relicensed my code [1] yesterday to a license I wrote [2], which is designed to poison the well for machine learning. [1]: https://gavinhoward.com/2021/07/poisoning-github-copilot-and... [2]: https://yzena.com/yzena-network-license/

Do the GitHub Terms of Service give them the necessary permissions for Copilot, independently of the license? (I honestly don't know the answer; this is a straight question.)

I don't know. Because I don't know is why I pulled all of my code (except for a permissively-licensed project that people actually depend on the GitHub link for) off of GitHub.

Re: All public GitHub code was used in training Copilot

#138

Post by Naga » Thu, Jul 08, 2021, 2:56 PM UTC

cool_dude85 wrote: post #110

jensensbutton wrote: post #98

Wouldn't it be the people publishing code written with Copilot that (potentially) violate any licenses? It doesn't seem to be that the tool violates anything, though it may put the _user_ at risk of violating something. Like, don't use it if you're worried about violating licenses, but I don't see how Microsoft could get in trouble for the tool. It doesn't write and publish code by itself.

Sorry, we built this tool for you that auto violates licenses. Sure, we're owned by a huge megacorp with billions of dollars, but it's your responsibility to confirm - and yes, we recognize it's impossible to confirm - that what you release using our tool isn't violating the license. In short, github gets to make the license violator bot and push the violations off onto the small fry who actually use it? No thanks.

Isn't that sortof the justification behind bittorrent and trackers?

Re: All public GitHub code was used in training Copilot

#139

Post by criddell » Thu, Jul 08, 2021, 2:58 PM UTC

diffeomorphism wrote: post #86

0x0 wrote: post #7

To the people arguing it's "fair use" to use this for training an ML network. Where do you draw the line? What if you train an "ML network" with one or two inputs... so that they almost always "generate" exact copies of the inputs? Five inputs..? Ten? A thousand? A million?

There obviously is no sharp line (e.g. it is 37. Immediate question: why not 36?), but that does not matter at all. We already have the same fuzzy line for writing. Am I forbidden from ever reading other author's books because I might accidentally "generate exact copies" of some of the sentences? Clearly not, that is how people learn a language. Does that mean I am allowed to copy the whole book? Also clearly not. Wh…

And somewhere is determined for your particular case in court. And tomorrow, a similar case may be determined differently.

Re: All public GitHub code was used in training Copilot

#140

Post by rubyist5eva » Thu, Jul 08, 2021, 2:58 PM UTC

Open source developers need a new kind of license with a ML model training clause, so there is no more ambiguity if they don't want their code to be used in this way.