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

NY Times copyright suit wants OpenAI to delete all GPT instances

arstechnica.com

Post reply on HN
Print view View original

501–510 of 921 posts

Re: NY Times copyright suit wants OpenAI to delete all GPT instances

#501

Post by edude03 »

munchinator wrote: post #180

It's interesting to me the ambiguous attitude people have to reproducing news content. Whenever there is a story from NYT on HN (or any other large media outlet), the top comment is almost always a link to an archived version which reproduces the text verbatim. And this seems to be tolerated as the norm. And yet, whenever there is a submission about a book, a TV show, a movie, a video game, an album, a comic book, or…

I think the intent is really different.

For LLMs you're essentially teaching them language by showing them lots of examples of written language - newspapers are of course a great example of written language.

The goal of OpenAI is not to reproduce newspaper articles verbatim when asked questions (even if the answer could be a newspaper article) and the fact that it can happen is a side effect of how LLMs work.

When a HN participant shares a (pay walled) link to a NYT article, I do want to read the exact article linked verbatim because while the facts of the article may be reproduced elsewhere in a form that's free, specific word choices or whatever might be a focal point of the discussion on HN, and therefore I can't realistically participate in a discussion without having read the article being discussed.

And as an aside, I have no problem with paying to read news, or whatever media, however it's impractical for me to subscribe to every news source HN participants link to, and therefore I gravitate to archiving services instead. I do wish there was a better solution - for example Blendle with more sources.

Re: NY Times copyright suit wants OpenAI to delete all GPT instances

#502

Post by logicchains »

andy99 wrote: post #492

This, or a lawsuit like it is going to be the SCO vs IBM of the 2020's, to wit: a copyright troll trying to extract rent, with various special interests cheering it on to try and promote their own agenda (ironically it was Microsoft that played that role with SCO). It's funny how times have changed and at least now a louder group seem to be on the troll's side. I hope to see some better analysis on the frivolity of t…

>It's funny how times have changed and at least now a louder group seem to be on the troll's side

Because for many people, their views on current events are whatever the "thought leaders" working for the NYT and similar publications tell them to think.

Re: NY Times copyright suit wants OpenAI to delete all GPT instances

#503

Post by phkahler »

andy99 wrote: post #492

This, or a lawsuit like it is going to be the SCO vs IBM of the 2020's, to wit: a copyright troll trying to extract rent, with various special interests cheering it on to try and promote their own agenda (ironically it was Microsoft that played that role with SCO). It's funny how times have changed and at least now a louder group seem to be on the troll's side. I hope to see some better analysis on the frivolity of t…

>> There may be some commercial subtlety in specific cases that doesn't depend on scraping and training

The key is to stop calling it "training" and use "learning" or just "reading".

The argument from NYT will probably be that LLMs are just a fancy way to compress or abstract information and spit it back out. In which case "training" seems to support their case?

Re: NY Times copyright suit wants OpenAI to delete all GPT instances

#504

Post by FrustratedMonky »

munchinator wrote: post #180

It's interesting to me the ambiguous attitude people have to reproducing news content. Whenever there is a story from NYT on HN (or any other large media outlet), the top comment is almost always a link to an archived version which reproduces the text verbatim. And this seems to be tolerated as the norm. And yet, whenever there is a submission about a book, a TV show, a movie, a video game, an album, a comic book, or…

Is this really copywrite?

Or is it "you can't talk to someone about an article they read".

This is really saying you can't call up your buddy and have them tell you a summary of what they just read. Maybe my buddy has a good memory and some of the text is actually nearly duplicate. But I wouldn't know because I didn't read the original, I just asked for a summary from someone else that read it.

Re: NY Times copyright suit wants OpenAI to delete all GPT instances

#505

Post by mattdesl »

bonzini wrote: post #468
Symmetry wrote: post #449

Earlier quoted context omitted.

In the screenshot for the article you can see that the LLM says it is "Searching for: carl zimmer article on the oldest DNA". That, and what I know about how LLMs work, suggest to me that rather than the article being stored inside the trained LLM it was instead downloaded in response to the question. So the fact that the system is providing the full text of the article doesn't really go to whether training the LLM i…

Yes, the screenshot in the article is clearly doing an Internet search. The exhibit in the lawsuit shows that you can complete an article by using GPT on the first sentence of the prompt, with low temperature to aid reproducibility, and obtaining the original except for a single word. That is another thing, and it shows that the LLM has basically recorded the original text into its weights in compressed form: https:/…

It would be curious to test this on a larger sample than just a few. It is hard to believe that a majority of NYT articles are verbatim stored in the weights of a web-wide LLM, but if that is the case it would be a pretty unbelievable revelation about their ability to compress an entire web’s worth of data. But, more likely, I assume it is a case of overfitting, or simply finding a prompt that happened to work well.

FWIW, I can’t replicate on either GPT 3.5 or 4, but it may be that OpenAI has added new measures to prevent this.

Re: NY Times copyright suit wants OpenAI to delete all GPT instances

#506

Post by chmod775 »

munchinator wrote: post #180

It's interesting to me the ambiguous attitude people have to reproducing news content. Whenever there is a story from NYT on HN (or any other large media outlet), the top comment is almost always a link to an archived version which reproduces the text verbatim. And this seems to be tolerated as the norm. And yet, whenever there is a submission about a book, a TV show, a movie, a video game, an album, a comic book, or…

>I think that's something worth reflecting on, about why we feel it's OK to pirate news articles, but not other IP.

A lot of of that is going to stem from the fact that respect for "journalism" is pretty low. More than 99% of news articles are copies of the <1% of original work that happens in that field. In news, everyone is already lifting content from everyone else.

Re: NY Times copyright suit wants OpenAI to delete all GPT instances

#507

Post by Mountain_Skies »

Looks like this is a case of Media vs Tech which might be solved by the courts using past paradigms but should really be addresses by legislation specific to this situation. The difficulty for the media companies, at least in the US, is that both major political parties see the media as the enemy. The left might be a bit more positive about the media but overall, they still see the media as something owned by wealthy elites suppressing knowledge of the harm the powerful inflict on the weak and powerless. Over on the Tech side of things, one party sees Tech as wholly owned by other side of the political divide. Over on that side, things are relatively (but not completely) friendly, so my guess is Tech will end up winning simply because it has more friends in the political realm than the Media does.

Re: NY Times copyright suit wants OpenAI to delete all GPT instances

#508

Post by rickydroll »

edude03 wrote: post #501
munchinator wrote: post #180

It's interesting to me the ambiguous attitude people have to reproducing news content. Whenever there is a story from NYT on HN (or any other large media outlet), the top comment is almost always a link to an archived version which reproduces the text verbatim. And this seems to be tolerated as the norm. And yet, whenever there is a submission about a book, a TV show, a movie, a video game, an album, a comic book, or…

I think the intent is really different. For LLMs you're essentially teaching them language by showing them lots of examples of written language - newspapers are of course a great example of written language. The goal of OpenAI is not to reproduce newspaper articles verbatim when asked questions (even if the answer could be a newspaper article) and the fact that it can happen is a side effect of how LLMs work. When a…

> The goal of OpenAI is not to reproduce newspaper articles verbatim when asked questions (even if the answer could be a newspaper article), and the fact that it can happen is a side effect of how LLMs work.

This is an excellent point. A properly functioning LLM should not return the original content it was trained on. When they return original content, I believe the prompt is tightly constrained and designed to extract or re-create original content. Another reason that occurred to me recently is that maybe the training set is too small, and more general prompts will re-create source material.

Another question would be, are LLMs regurgitating what they were trained on, or are they synthesizing something very close to the original content? (Infinite Monkeys, Shakespeare). Court cases like this increase the need for understanding the "thinking processes" in an LLM.

Re: NY Times copyright suit wants OpenAI to delete all GPT instances

#509

Post by DennisP »

Matticus_Rex wrote: post #487
DennisP wrote: post #473

Earlier quoted context omitted.

I wouldn't say OpenAI has exactly the same attitude, since they also pulled in thousands of books. Their position has been that it's not piracy, since they don't republish the books; effectively the AI just reads them and learns from them. If GPT can be made to reproduce the original articles, that's a more difficult argument to make.

It turns out you can reproduce articles with next-token prediction when the articles are quoted all over the dataset. The articles themselves are indisputably not a part of the model, because it doesn't store text at all. OpenAI's position is correct; people just underestimated how well the AI learns from reading, especially when it reads the same text in a bunch of different places because it's being quoted/excerpte…

What seems a bit contradictory is that they're also suing because GPT hallucinates about NYTimes articles. So they're complaining that it reproduces articles exactly but also that it doesn't.

Re: NY Times copyright suit wants OpenAI to delete all GPT instances

#510

Post by vidarh »

cmiles74 wrote: post #423
vidarh wrote: post #183

Earlier quoted context omitted.

The NYT is also worth a tiny fraction of that. If it looks like they might get anywhere, it might be better for OpenAI to buy them

OMG! Or they could just license the content. I suspect that would be both easier and less expensive. ;-)

I'm not convinced it's a given it will. If it becomes necessary to license, owning the large publishers will be leverage and allow locking competitors out unless you have a portfolio to cross license.

OpenAI alone has a market cap that'd allow it to buy about as large a proportion of publishers of newspapers and books as they'd be allowed before competition watchdogs will start refusing consent.

Put another way:

If I was a VC with deep pockets investing in AI at this point, I'd hedge by starting to buy strategic stakes in media companies.

Post reply on HN