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

881–890 of 1001 posts

Re: GitHub Copi­lot inves­ti­ga­tion

#881

Post by ghaff »

adamhp wrote: post #843
bayindirh wrote: post #771

Earlier quoted context omitted.

> Microsoft will likely need to add checks to prevent copilot from offering verbatim copies of code going forward to try to avoid copyright violations here. A user once replied to one of my comment[0] about this with the following: > It's not really an issue when you're a large software corporation; you already have mechanisms in place to check for license compliance in everything that ships, including F/OSS plagiari…

I'm curious about the endgame of copyright with respect to software. At some point, enough people will have written enough code that you can't write code anymore because some fragment of it violates a copyright. Where does the line get drawn? There's only so many ways to do certain algorithms, like DFS or BFS.

>you can't write code anymore because some fragment of it violates a copyright

Copyright (unlike patents in general) allows for independent creation. If I sit down to write a quicksort routine, it is going to look extremely similar to a zillion other quicksort routines out there.

The other question (IANAL) is whether writing a quicksort routine is even a creative act at this point.

Re: GitHub Copi­lot inves­ti­ga­tion

#882

Post by mahogany »

throwaway09223 wrote: post #429
mahogany wrote: post #374

Earlier quoted context omitted.

> 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. Following this line of thought, do you think that all code from all software should be open source and publicly available (and free to copy and use), in the…

> Following this line of thought, do you think that all code from all software should be open source and publicly available Let's help shape this thought: Copyright should be abolished entirely. It is one of many monetization schemes and its negative effects greatly outweigh its positives. We know people won't stop writing software in the absence of copyright. We know they won't stop writing books, singing songs, etc…

> Let's help shape this thought: Copyright should be abolished entirely. It is one of many monetization schemes and its negative effects greatly outweigh its positives.

Even if you're right in principle (and I would love new monetization structures), this will never happen in reality.

Meanwhile, this idealism will get applied asymmetrically in the real world. If you (or the comment I was replying to) say "Copilot is fine, all code should be publicly available anyway", it downplays the fact that this wish will never happen with big players like Microsoft and will only happen with little players like anyone who used Github to host their code. The big player will typically hide their code behind copyright and lawyers to enforce it, whereas the little players have no similar recourse.

So, I see the issue as an exploitation, as Microsoft is selling a product built on the little players and not the big players. The debate around whether copyright should exist at all, while interesting, is not that relevant to most of the concerns being aired in the context of Copilot.

Re: GitHub Copi­lot inves­ti­ga­tion

#883

Post by vasco »

i_like_apis wrote: post #821
BowBun wrote: post #794

I'm in favor of this. You can't ingest code that says "you cannot use this without attribution", put it through a bunch of if statements that strip the license, and then say it's "AI-generated". I don't care about most of our generic CRUD apps or the 15th rewrite of a sorting algorithm, but I do care about those smart enough to advance the field and come up with novel solutions. If we take away the incentive for attr…

Totally disagree. Training is fair use. It is akin to learning. Code licenses do not restrict you from reading or learning. ML training needs to be fair use of copyrighted works, or most machine learning and AI projects will be impossible.

Microsoft has written and acquired plenty of codebases over the years. Train the AI on that.

Re: GitHub Copi­lot inves­ti­ga­tion

#884

Post by _dongle_aster_ »

kmeisthax wrote: post #568

If Copilot itself is infringing then so is GPT-3, DALL-E 2, NovelAI, and Stable Diffusion. There's no legal argument that would solely target one application of this technology, and you can't build generative AI using current ML tools without relying on a very large corpus of public data. All AI is built on free-riding[0]. While there is no US case law that explicitly says "training AI is fair use", the Second Circui…

Are the examples of stable diffusion exactly reproducing images from its training set?

Re: GitHub Copi­lot inves­ti­ga­tion

#885

Post by macspoofing »

woah wrote: post #491

It would be sad if someone succeeded in shutting down CoPilot for this kind of copyright stuff. It is genuinely useful. I don't care that it reproduces copyrighted content. The only way you can get it to do that is to bait it with the function names of functions that have already been copy and pasted thousands of times onto GitHub without proper licenses. Luckily, someone will probably come out with a "renegade" vers…

>The only way you can get it to do that is to bait it with the function names of functions that have already been copy and pasted thousands of times onto GitHub without proper licenses.

How do you know that this is the only way for copilot to reproduce copyrighted code?

Re: GitHub Copi­lot inves­ti­ga­tion

#886

Post by treis »

sushid wrote: post #302
treis wrote: post #190

Earlier quoted context omitted.

You're not supposed to be able to use dominance in one market (git hosting) to gain dominance in another (AI powered code suggestions).

You might be confused but that's literally what you do as a business. You leverage your domain area expertise to expand into new areas of business. For instance, Apple already knew about the portable hardware market and extended their reach into the portable music market via iPod. It used the iPod to reach the music marketplace via iTunes. Used its market dominance to create iPhone and the rest is history. Maybe that…

>You might be confused but that's literally what you do as a business. You leverage your domain area expertise to expand into new areas of business.

>or instance, Apple already knew about the portable hardware market and extended their reach into the portable music market via iPod. It used the iPod to reach the music marketplace via iTunes. Used its market dominance to create iPhone and the rest is history.

Look at what you're saying here. Apple had domain experience in hardware and launched a new product (good). Then it used the dominance of that product to muscle into an entirely different market (bad). And the combination of hardware and market has led to Apple being able to extract their tax on half the music market, or whatever they have.

This is exactly what we don't want and why anti-trust laws exist.

Re: GitHub Copi­lot inves­ti­ga­tion

#887

Post by i_like_apis »

BowBun wrote: post #872
i_like_apis wrote: post #821

Earlier quoted context omitted.

Totally disagree. Training is fair use. It is akin to learning. Code licenses do not restrict you from reading or learning. ML training needs to be fair use of copyrighted works, or most machine learning and AI projects will be impossible.

Like posts have said, whether training is fair use is not a matter of opinion, it is a matter of law. You can't use an appeal to your authority to make grand statements like this. Frankly, I don't care that ML/AI _needs_ this to work. That's not my problem. You don't get to circumvent existing agreements (and law) because you believe that ML learning is the same as a human reading a piece of code and then typing it u…

There isn’t law any law yet.

And yes, it is the same thing as learning.

If you have a robot that learns like a human does … you think it should be illegal for that machine to look at GitHub? To watch a Hollywood movie?

Re: GitHub Copi­lot inves­ti­ga­tion

#888

Post by LtWorf »

citizenkeen wrote: post #109
armchairhacker wrote: post #37

One issue I see with Copilot is that they get free access to all open-source data on GitHub, but using GitHub APIs to download the data yourself isn't possible (rate limiting). This is an unfair advantage. Copilot is not only making money off of open-source, they are making money off of open-source in a way others can't. I would love to see a lawsuit which requires GitHub to provide their full Copilot dataset.

It’s not making money off of open source, it’s making money off of hosting open source.

No? How much money would it make without using open source?

Re: GitHub Copi­lot inves­ti­ga­tion

#889

Post by burkaman »

sweetheart wrote: post #583
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.

I don’t think anyone really cares about that kind of stuff, though. Like, there are loads of examples of similar things which no one (rightfully) bags an eye at. Like if I take a photograph of you out in the public and sell the photo for $1 million, would you expect compensation? Or if someone compiles a list of the best restaurants in the world and sells that list, do you think the restaurants should be compensated?…

> if I take a photograph of you out in the public and sell the photo for $1 million, would you expect compensation?

I mean I wouldn't expect it, but I think I'd be pretty annoyed if you didn't ask permission and then made a bunch of money off my image. It's easy to find stories from the subjects of famous photographs who feel like they've been exploited. Just off the top of my head there's Afghan Girl, the kid from the Nirvana album, Harvard's collection of photos of enslaved people, and Henrietta Lacks is sort of a similar case.

> if someone compiles a list of the best restaurants in the world and sells that list, do you think the restaurants should be compensated?

No, but here's a better example: you make friends with a bunch of food critics, collect their thoughts and opinions and favorite secret spots, and then publish a book based on that stuff without ever telling them what you were doing or compensating or crediting them.

I'll give a concrete example: I was rock climbing recently and met an old guy who was sort of the local expert, and he told me how some other non-locals had come in and kind of mined him for information about the area, all the routes, etc. and then published a guidebook without crediting him at all. He felt pretty upset and exploited by that, and I felt bad for buying the guidebook because I had assumed it was written by some local climbers and didn't realize they got most of their info from someone else.

It's not illegal, but it is unethical.

Re: GitHub Copi­lot inves­ti­ga­tion

#890

Post by sfpotter »

deworms wrote: post #157
lilyball wrote: post #55

Earlier quoted context omitted.

And would therefore have to follow the license of the code they took it from. That's exactly the point. Copilot is reproducing the same code but without the license.

A simple search on github reveals that those functions were reposted verbatim thousands of times, most people just copy and paste snippets of code they find useful, ignoring licenses. This highlights how all the power a license promises to hold is completely fictional. Any "in the style of Tim Davis" modifier only shows some kind of unwarranted self-importance complex on the part of the guy, thinking his style is wid…

It does not show an unwarranted sense of self-importance on the part of Tim Davis.

Whether or not his style his widely known, his code is VERY widely used. Just look up SuiteSparse and try to find all of the downstream uses of it. It is one of the most---if not the most---ubiquitously used set of sparse linear algebra libraries. If you do anything with numerical linear algebra, there's a good chance you at least know what SuiteSparse is, and possibly also know who Tim Davis is.

The bigger issue here is the effect this has on research. Tim Davis not only programmed this library, he did the basic research leading to many of the algorithms in SuiteSparse. He went ahead and released SuiteSparse open source, probably thinking that it would be a good deal for him, provided that its use was properly attributed. Provide a public service in exchange for attribution. This is a reasonable way to get support as an academic. Clearly he has had a large number of industrial collaborations which likely have provided him with a significant amount of funding over the years.

Speaking for myself, if Microsoft has no compunction against behaving this way, I can no longer see the point in publicly releasing research code that I develop using an open source model. Microsoft is clearly telegraphing that they don't give a f** about licensing, although whether that holds if they are litigated against remains to be seen. I think there's an excellent chance many other researchers feel the same way. If you think openness and reproducibility in science is important, this is a problem.

Post reply on HN