41–50 of 734 posts
OP’s rhetoric, and most discussion I see, asserts that training a model on copyrighted data is a copyright violation. Personally I don’t find this to be so obviously the case. Think back to when we were listening to AI generated pop music, for instance. I don’t recall any concern in HN comments about the copyright holders’ music being used for learning.
Out of curiosity, how do we define license violation in that case? I, as a human being, have trained by reading code, much of which is covered by licenses that are somehow not compatible with code I'm writing. Am I violating licenses? Asking seriously. It's really unclear to me where law and/or ethics put the boundaries. Also, I'd guess it's probably country dependent.
> Out of curiosity, how do we define license violation in that case? I, as a human being, have trained by reading code, much of which is covered by licenses that are somehow not compatible with code I'm writing. Am I violating licenses? That depends, if you end up writing copies of the code you've studied then yes. You are on thin ice. Plagarization is definitely something that you can do with computer code. There ha…
As for fronted/open source etc... sure, if you don't care about copyright and licensing, use it.
As this boils down to legal arguments, are there any clauses (maybe disputed) in the ToS allowing github/MS usage of public repos for such purpose?
Would it even be legally possible to override a software license as a repo provider like "by using this service, you agree to..."?
Out of curiosity, how do we define license violation in that case? I, as a human being, have trained by reading code, much of which is covered by licenses that are somehow not compatible with code I'm writing. Am I violating licenses? Asking seriously. It's really unclear to me where law and/or ethics put the boundaries. Also, I'd guess it's probably country dependent.
The training question seems much more difficult. The main problem that has been the topic is a simpler one - about the produced work. If you exactly reproduce someone's existing code (doesn't matter if you copy by flipping bits one by one or which technology you use), isn't it a copyright violation? I'm kind of imagining a Rube Goldberg machine that spells out the quake invsqrt function in the sand, now...
With programming, there's the further complication what constitutes a work. But quakes invsqrt certainly qualifies, just like that one function from the Oracle vs Google case.
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 press a button, it inserts the line from the array, and press it again and you get the next line, etc.
Now hopefully nobody here thinks such a system would fulfil either the spirit or the law of any half-restrictive license. Yet that is a perfectly valid implementation of Copilot's aim - and it sounds like it's not that far from what actually happens, maybe with a bit of variable name munging.
So my question is this: If you could build a line between the system I describe above and the system of human learning, where a human learns the patterns and can genuinely produce novel structures and patterns and even programming languages that it has never seen before.
At what point along that line would you say that Copilot is close enough to human to not be violating licenses that require attribution?
I don’t know how to approach this. As a human I can read all public code and learn from regardless the license and come up with new solutions. Machine can read everything too, but can't create new ideas or approach. How is copilot defined then? Should it be only smart system for general code snippets?
Earlier quoted context omitted.
Indeed; the question is if copyright should apply at all . Harping on about licenses, GPL, and whatnot is a detraction from the actual issue at hand. Also, given that the author of this tweet called me a "bootlicker" last year in response to a somewhat lengthy nuanced post about GitHub, I'm gonna go out on a limb and say that they're not all that interested in a meaningful conversation on this in the first place but…
The possibility of GPL violation does show (one of) enormous ramifications of the question though. I think it's not a detraction as long as the question itself is also mentioned.
I don’t know how to approach this. As a human I can read all public code and learn from regardless the license and come up with new solutions. Machine can read everything too, but can't create new ideas or approach. How is copilot defined then? Should it be only smart system for general code snippets?
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.
OP’s rhetoric, and most discussion I see, asserts that training a model on copyrighted data is a copyright violation. Personally I don’t find this to be so obviously the case. Think back to when we were listening to AI generated pop music, for instance. I don’t recall any concern in HN comments about the copyright holders’ music being used for learning.
If one of the three largest record labels uses their own catalog to train an AI, copyright seems less important to discuss. I suspect the discussion would be a bit different if a company scraped youtube and used that as a training set for AI music and successfully sold it.