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

AI is just unauthorised plagiarism at a bigger scale

axelk.ee

Post reply on HN
Print view View original

551–560 of 783 posts

Re: AI is just unauthorised plagiarism at a bigger scale

#551

Post by Someone »

jongjong wrote: post #444
danorama wrote: post #414

There’s a fallacy that gets used a whole lot to justify things like this (not just with LLMs), and I see it in many of the comments here: If it’s OK (or at least negligible on a small scale), then it must be OK on a large scale. It usually goes something like: If I can make money by learning something from a web page, why does a computer making money by learning everything from everyone upset people so? It’s the same…

This is a great point. I think for coding, the wording of the MIT open source license makes it clear that copying and distributing the software is authorised on a small scale and it's very clear that the act of copying must involve a person. It provides distribution and modification rights to "any person obtaining a copy of the software" and explicitly requires attribution for any significant parts. Mass-ingesting th…

> I think for coding, the wording of the MIT open source license makes it clear that copying and distributing the software is authorised on a small scale and it's very clear that the act of copying must involve a person.

I agree with “must involve a person. https://opensource.org/license/mit starts with (emphasis added) “Permission is hereby granted, free of charge, to any PERSON obtaining a copy of this software and associated documentation files (the “Software”)”.

That means it doesn’t give an LLM any rights. The way I see it, LLMs run (directly or indirectly) by a person can do stuff on their behalf, though, just as your CI pipeline can download and compile MIT-licensed software.

I definitely disagree with the “on a small scale” as the license continues (again, emphasis added) “to deal in the Software WITHOUT RESTRICTION, including WITHOUT LIMITATION the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software”.

Re: AI is just unauthorised plagiarism at a bigger scale

#553

Post by nonethewiser »

deaton wrote: post #30

"Steal an apple and you're a thief. Steal a kingdom and you're a statesman." - Literal Disney villain

Yeah but isn't the claim that it's not stealing? You're presupposing the thing that is in question.

The argument is that a human will gather information from all over the place and compile it, all without doing anything wrong. That's the base claim. Not that stealing a little is OK. That's extremely easy to disprove and also entirely irrelevant.

Re: AI is just unauthorised plagiarism at a bigger scale

#554

Post by 6031769 »

vitally3643 wrote: post #449
wolttam wrote: post #420

Earlier quoted context omitted.

Yes, but: > Although Anubis could be altered to mine cryptocurrency to serve as proof of work, Iaso has rejected this idea: "I don't want to touch cryptocurrency with a 20 foot pole." Which in my mind is a shame. Crypto is an absolute mess, yes, but this seems like an elegant way to get something back for putting things out there.

Mining crypro doesn't materialize money. You have to exchange it for real money which means taking a private individual's money in exchange for scam tokens. This is the problem crypto fans refuse to acknowledge. The money doesn't magically appear, you're taking it from someone else and letting them hold the bag when whatever cryptocurrency you choose inevitably blows up, fails, or rug-pulls. It's unethical to engage…

Not necessarily. You can spend your cryptocoins with any number of businesses and it is very much the choice of those businesses to accept them or not. No private individuals need be involved.

Note also that any non-crypto currency can also devalue at any moment, although perhaps not to the same extent. Holding anything of any perceived value carries a risk and also a potential reward.

Re: AI is just unauthorised plagiarism at a bigger scale

#555

Post by 4chandaily »

mitthrowaway2 wrote: post #455
4chandaily wrote: post #418

Earlier quoted context omitted.

You may want to review your history. The GPL is copyleft -it only exists to subvert copyright law by using it against itself in a sort of intellectual legal judo. If "IP" laws were not as they were, there would be no need for the GPL. Software would be Free. https://en.wikipedia.org/wiki/Copyleft

Even if companies didn't have copyright protection on their source code, that doesn't mean they'd post it all on the internet for anybody to freely download.

No, not all of them, but some companies, many organizations, and plenty of individuals would.

Not everything has to be done for a profit. Plenty of us make software, art, and technology because we find it fun and interesting to work on, and because we want to live in a world that is richer for it.

Removing draconian intellectual property laws that mostly only benefit the giant corporations that lobbied for them isn't going to stop me from doing so, and I doubt it would stop many others.

Re: AI is just unauthorised plagiarism at a bigger scale

#556

Post by nonethewiser »

rib3ye wrote: post #520
deaton wrote: post #30

"Steal an apple and you're a thief. Steal a kingdom and you're a statesman." - Literal Disney villain

If you agree with this, then you have to question the validity of international "law"

Do you not question the validity of international law? How can something be a law over sovereign countries without some overarching governing body and enforcement mechanisms? How can it be a law if the parties it supposedly governs can just recognize or not recognize it at will?

Re: AI is just unauthorised plagiarism at a bigger scale

#557

Post by mullingitover »

danorama wrote: post #414

There’s a fallacy that gets used a whole lot to justify things like this (not just with LLMs), and I see it in many of the comments here: If it’s OK (or at least negligible on a small scale), then it must be OK on a large scale. It usually goes something like: If I can make money by learning something from a web page, why does a computer making money by learning everything from everyone upset people so? It’s the same…

> why does a computer making money by learning everything from everyone upset people so? It’s the same thing!

The majority of the population, sitting outside the VC bubble, views AI unfavorably. That's not my hot take, that's a fact from the NYT survey published today.

It's going to be hilarious when VCs, having expropriated the IP of the entire internet, build The Layoff Machine That Does Everything Without Workers, and then the voters decide to just...enthusiastically expropriate that, and we end up with Fully Automated Luxury Communism.

Re: AI is just unauthorised plagiarism at a bigger scale

#558

Post by Meph504 »

danorama wrote: post #414

There’s a fallacy that gets used a whole lot to justify things like this (not just with LLMs), and I see it in many of the comments here: If it’s OK (or at least negligible on a small scale), then it must be OK on a large scale. It usually goes something like: If I can make money by learning something from a web page, why does a computer making money by learning everything from everyone upset people so? It’s the same…

> It’s like if I go to Golden Gate Park and pick one flower, I shouldn’t do that, but no one cares. But if I build a machine to automatically cut every flower in the park because I want to sell them, that’s different.

The problem here is, in your example the small scale example, and the large scale example are both unacceptable behavior.

Learning from others at a small scale is not only socially acceptable, but is the foundation of how advancement works.

So this concept of the issue of the scale being the issue isn't at its core the problem, its that something that that is desired behavior in a human, is not socially acceptable because of a machine is doing it.

Re: AI is just unauthorised plagiarism at a bigger scale

#559

Post by bigbuppo »

jhbadger wrote: post #522
disposition2 wrote: post #511

Earlier quoted context omitted.

You could also easily opt out with the single entity that shipped that information.

Yes, but getting an unlisted number was considered weird and against the norm even if possible. Even in the early 2000s when I dropped my landline, my parents were aghast - "if you do that, you won't be in the phone book! How will anyone get in contact with you?"

And you often had to pay for the privilege... A dollar a month for them to not put your name and number in the phonebook.

Re: AI is just unauthorised plagiarism at a bigger scale

#560

Post by nonethewiser »

20k wrote: post #540
lbrito wrote: post #245

Earlier quoted context omitted.

>You are never going to get the exact book word-for-word using LLM. This is pretty much the exact claim of a NYT lawsuit against OpenAI. "One example: Bing Chat copied all but two of the first 396 words of its 2023 article “The Secrets Hamas knew about Israel’s Military.” An exhibit showed 100 other situations in which OpenAI’s GPT was trained on and memorized articles from The Times, with word-for-word copying in re…

Yes, LLMs fundamentally operate as a lossy compression scheme for their training data. There's been countless examples of them reproducing their training data with very high accuracy People claim that the data isn't stored, but clearly a representation of it is encoded and reproducible. I saw chatgpt word for word plagiarise a stack overflow comment just two days ago

Does this actually imply a representation of it has been stored or simply that the model is sort of over-fit?
Post reply on HN