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

Ownership of AI-Generated Code Hotly Disputed

spectrum.ieee.org

Post reply on HN
Print view View original

101–110 of 205 posts

Re: Ownership of AI-Generated Code Hotly Disputed

#101

Post by mellosouls »

megous wrote: post #18
mellosouls wrote: post #12

I would have thought that in the vast majority of current AI-generated code we are talking about single blocks and functions that are just Intellisense on steroids that only a rather self-deluding coder would consider original enough or "theirs" to attribute authorship to. There are no doubt grey areas and more serious cases as the technology improves and the generated content increases in length and functional value…

You can generate whole linux kernel drivers with eg. ChatGPT. And things will only get better, so your dismissal is not very forward looking. The issue will need to be dealt with.

You should reread my comment to see that I explicitly made exceptions for more advanced cases.

Re: Ownership of AI-Generated Code Hotly Disputed

#102

Post by TchoBeer »

thn-gap wrote: post #79
c7b wrote: post #22

Earlier quoted context omitted.

Even that notion is tricky in the context of text AI. It doesn't seem obviously clear to me what the implied stance towards AI training would be for many popular license types. It seems more like the creators of many licenses didn't explicitly consider this use case. And the more meta question is whether the creators' rights should even extend to that realm. Can you specify in a license "This text must not be read by…

Why is it fair use if you train a model on copyrighted material and use its transformative output, despite going against the will of the author? Can we all legally pirate educational books since it's for self training and producing transformative outputs? Can I consume all media (books, movies, music) the same way, and call it fair use? Also for software?

yeah, I struggle to see how training AI on illegally gotten data is legal.

Re: Ownership of AI-Generated Code Hotly Disputed

#103

Post by Rekksu »

falcolas wrote: post #39
wrs wrote: post #16

“…modify[ing] its AI model so that it traces attribution and gives credit to the original authors of the code, adding the associated copyright notices and license terms in the process…Biderman says is technologically feasible.” Is it really feasible? What does “traces attribution” even mean here? It’s not emitting “code”, it’s emitting individual tokens that each were found throughout the input corpus. The “code” is…

> It’s not emitting “code”, it’s emitting individual tokens ... The “code” is the arrangement of those tokens, but that is determined by the weighting of the whole network This theory of operation is not borne out in reality. It's been clearly displayed that these tools are emitting verbatim copies of existing code (and its comments) in their input. It's even being seen in image generation, where NatGeo cover images…

Do you have a source for this claim? This isn't really how generative models work

Re: Ownership of AI-Generated Code Hotly Disputed

#104

Post by falcolas »

safety1st wrote: post #93
falcolas wrote: post #39

Earlier quoted context omitted.

> It’s not emitting “code”, it’s emitting individual tokens ... The “code” is the arrangement of those tokens, but that is determined by the weighting of the whole network This theory of operation is not borne out in reality. It's been clearly displayed that these tools are emitting verbatim copies of existing code (and its comments) in their input. It's even being seen in image generation, where NatGeo cover images…

Even if that theory of operation were true, it seems like an implementation detail that would be largely irrelevant in a court of law. Infringement is infringement regardless of how complicated your method for committing it was. If your AI reproduces a copyrighted work verbatim without permission, surely you are committing infringement? If you would be infringing by doing it via copy and paste, I would think a court…

Personally, I agree with you. Others do not.

But irrespective of legality, my point is that attribution is not such a hard problem as people believe, because their mental model of what's going on is much more complicated than reality. I feel they are seeing it as a true AI (mostly non-deterministic and thus hard to observe) and not ML which is deterministic.

Re: Ownership of AI-Generated Code Hotly Disputed

#105

Post by mellosouls »

vouaobrasil wrote: post #15
mellosouls wrote: post #12

I would have thought that in the vast majority of current AI-generated code we are talking about single blocks and functions that are just Intellisense on steroids that only a rather self-deluding coder would consider original enough or "theirs" to attribute authorship to. There are no doubt grey areas and more serious cases as the technology improves and the generated content increases in length and functional value…

I think that kind of depends on how common the snippet is. If it's code for swapping to variables, well everyone has written that and it's small. But what if it's a rarer thing like a new kind of debayering algorithm that takes a much larger "snippet" from a novel algorithm? As far as I'm aware there are no rigorous studies showing the level to which one could duplicate longer stretches of code and if it can do somet…

Yes, of course, but that isn't the "vast majority" of use cases I pointed out; most people will presumably be using it for bog standard stuff in place of googling and thinking for a short while - let alone the people who didn't need the memory jog and are just selecting from an automatic prompt to save them typing it.

Re: Ownership of AI-Generated Code Hotly Disputed

#106

Post by dahart »

wrs wrote: post #71
dahart wrote: post #54

Earlier quoted context omitted.

Why wouldn’t it be feasible? (Maybe this depends on what you mean by ‘feasible’.) There’s no technical reason you can’t back-track the weights and make a list of which tokens from which training data were sampled. The list might be long, it could be impractical, but that has little bearing on whether it’s technically possible, right? The problem here happens when the same source is sampled for many tokens in a row be…

The token “if” appears in a heck of a lot of inputs, and a heck of a lot of outputs. What you’re describing seems like basically building an inverted index of the input corpus and then doing a search for the output. If the answer comes back with a high relevance then you consider it “traced”. I wonder what the size of that inverted index would be compared to the size of the generative model.

That sounds exactly right to me, or at least this is one specific way to implement a solution to the problem. Good question on size. Speculating… I would guess the size of the index relates not to the size of the model, but to the size of the training data. If you built an extremely naive and straightforward uncompressed index like this, you could imagine building the complete list of known tokens, and for each one the complete list of pairs of source index and character offset into each source. If a pair of 64 bit ints was sufficient for this, you’d have 128 bits per token for the index, compared to whatever the average size token is (which I’d guess is something like 2-4 bytes uncompressed). If that napkin math is anything close to reality, then I might expect the index to be ~8x the size of the training data (with error bars of like 2x-4x on both sides). Maybe we just call it an order of magnitude?

Re: Ownership of AI-Generated Code Hotly Disputed

#107

Post by somrand0 »

claytongulick wrote: post #89
somrand0 wrote: post #2

ownership is a concept in dire need of revision

Agreed. You don't mind if I borrow your car do you? And your house? Let's not quibble about the particulars of me ever giving them back.

indeed, the first step (IMO) towards revising the notion is to recognize that physical (material, tangible) assets inherently work differently than digital assets.

As I understand so far the main reason to seek a revision of the concept of ownership is exactly due to the existence (enabled by internet technology) of digital assets.

copy-pasting is HOW computers work. copy-pasting does not do well in society ruled by the exclusivity-mindset inherent to marketplaces (and their societies) of tangible assets

Re: Ownership of AI-Generated Code Hotly Disputed

#108

Post by falcolas »

Filligree wrote: post #83
falcolas wrote: post #66

Earlier quoted context omitted.

https://youtu.be/kqPKNksl9hk?t=80 https://news.ycombinator.com/item?id=33061707 <- Watermarks (there's been a few linked here on HN)

The AI learns to generate watermark-like things, because those exist on a large fraction of its inputs. It doesn't mean the rest of the output existed in the training set. It's perfectly capable of generating a completely novel picture, then slapping a watermark on it.

My point is that the watermark is deterministic. It is taken whole-cloth from an input, and reproduced as-is on an output. Thus, it is able to be attributed.

Stable diffusion is more like a paintshop artist who grabs bits and pieces of other art and melds them together, and less like a painter who creates from their imagination.

Re: Ownership of AI-Generated Code Hotly Disputed

#109

Post by ProlificInquiry »

wrs wrote: post #16

“…modify[ing] its AI model so that it traces attribution and gives credit to the original authors of the code, adding the associated copyright notices and license terms in the process…Biderman says is technologically feasible.” Is it really feasible? What does “traces attribution” even mean here? It’s not emitting “code”, it’s emitting individual tokens that each were found throughout the input corpus. The “code” is…

This quote seems to fundamentally misunderstand what transformers are doing at all. Technically I suppose you could save all gradient updates from every input token, and do some weighted averaging to show what inputs affected the particular output the most, but saving all those gradient updates would be unimaginably space consuming. "Feasible" is doing a lot of work there.

It's very hard for people to get away from the idea that GPT is "copying" something, but that's not what it's doing. The reality is, to get the exact artifact which produced the code in question, you need "Call me Ishmael" from Moby Dick just as much as the Linux kernel source.

Re: Ownership of AI-Generated Code Hotly Disputed

#110

Post by stale2002 »

dbspin wrote: post #40
aftbit wrote: post #33

Earlier quoted context omitted.

>Many artists the world over are rightfully furious about DALL-E and Stable Diffusion I struggle with this one. How are these models different from the typical human creative process of: 1. look at lots of existing art to get inspiration 2. select components from several different styles & add your own flair 3. call the output an "original" painting in your own style Of course I see the other side as well. These mode…

In a word scale. Scale matters. How is one locust different to a million? How is hand copying a manuscript (highly controlled in medieval Europe) different to a printing press. These models will and are already having a profound economic impact on the creative sectors they mimic. Whether existing law applies is a specialist question. But they're not remotely comparable from a pragmatic perspective with pieces manuall…

> In a word scale. Scale matters

> to shrug while others lose their livelihood and have their self expression systematically duplicated and commodified.

The same exact arguments were made when the camera was invented.

In order to be consistent, you would have to argue that the camera should have been banned to protect painters.

> That you can't empathise or find it 'naive'

Oh we understand. We just know that it is the exact same argument, that ludites make every single time a new technology is invented, that increases efficiency.

It is always the same argument repeated over and over again. And if those arguments weren't good in the past, we aren't going to listen to defeated arguments now.

Post reply on HN