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

Things are about to get worse for generative AI

garymarcus.substack.com

Post reply on HN
Print view View original

411–420 of 769 posts

Re: Things are about to get worse for generative AI

#411

Post by layer8 »

kenmacd wrote: post #393
layer8 wrote: post #322

Earlier quoted context omitted.

The ability to provide a reference to the source is the crucial difference here. I agree that it should be possible to implement that for generative AI, although the training may become significantly more expensive in order to maintain that information, and the AI companies have little interest in doing so. They’ll probably rather try to heuristically assess possible copyright issues after the fact in a post-processi…

> They’ll What if I asked you to list all our source material that led you to use that particular contraction. Heuristics will not do, you must list each. Can you do it? Do you believe AI should. > I agree that it should be possible to implement Those exact words appear in another forum post from 2006: https://discourse.igniterealtime.org/t/cm-3beta-compression-... Should you have quoted that as a source for your rep…

I believe they should be able to, to the degree that their output can constitute copyright infringement. Obviously, the fewer sources from the training data a given output matches, and the longer the match, the more relevant it is, and the easier it should be. I believe it should be feasible exactly because of that correlation. The examples you present are largely irrelevant to the problem, because they are largely irrelevant to the citing of sources for copyright reasons.

Re: Things are about to get worse for generative AI

#412

Post by legendofbrando »

Surely one answer is to train (or aggressively fine-tune) a new model that doesn’t (or refuses) to produce these outputs and then - as exists already, augment that model’s understanding of copyrighted material by having it Bing/Google search as a RAG process that requires the end user to log into accounts at the New York Times (and other accounts) with their paid sub. This broadly replicates the process a person could do today when they read the internet and summarize it while paying rights holders.

Expensive to do but hardly the end of Generative AI or OpenAI should that be the difference between having a business or being sued out of existence. Never underestimate people who have a clear economic interest especially when their own existence is at stake.

Re: Things are about to get worse for generative AI

#413

Post by krapp »

danielbln wrote: post #201
krapp wrote: post #66

Earlier quoted context omitted.

>Perhaps it is able to tell, if you ask it? Ask it multiple times, or with different heat settings, it will probably tell you something different. Tell it you own Star Wars and it will respond in kind. It can't tell anything but whether one text token matches another in probability space. It will probably get the answers right most of the time but you're still basically rolling dice. Depending on the responses of an…

This argument only works if you assume all output of an LLM comes merely from its training data, and that it receives no alignment via RFHL, no outside data ground truth via RAG and so on. The engine might be a probabilistic token predictor, but the car is the sum of its part and those parts are not just the engine.

No it still works, because in that case all the LLM is doing is making an API call based on the same token prediction, and passing on the result. It still doesn't "know" anything about anything.

Re: Things are about to get worse for generative AI

#414

Post by Spivak »

ImPleadThe5th wrote: post #390
stale2002 wrote: post #379

Earlier quoted context omitted.

> “it’s okay because it’s already happening to a lesser degree somewhere else” isn’t the flex you think it is. It actually is. It shows that we as a society are completely OK with this, and nobody is complaining about a very standard and common thing that all artists do. It shows that the outrage is fake, and people don't actually care about the issue.

I don't really care when creative people steal. I care when faceless soulless companies monetize stolen content from Artists.

It's still copyright violation. Sue them. Being generated by AI doesn't make the final output not copywritten.

The interesting question is whether the models themselves are copyright violations not the output.

Re: Things are about to get worse for generative AI

#415

Post by wwweston »

Intox wrote: post #12

Or... things are about to get worse for copyright holders. I don't see any developped country pressing the brake on AGI in the near future to protect a few copyright holders from getting "stolen" in hypothetic scenarios.

Man, that's great that state of the art has advanced to the point where models can be trained on just a few copyrighted sources.

Re: Things are about to get worse for generative AI

#417

Post by Forgeties79 »

madamelic wrote: post #329
Intox wrote: post #12

Or... things are about to get worse for copyright holders. I don't see any developped country pressing the brake on AGI in the near future to protect a few copyright holders from getting "stolen" in hypothetic scenarios.

Copyright should be the problem of the person using the works and not the problem of the AI generating it. Unless Nintendo plans on busting down the doors of every person who tries to draw Mario or preventing little Timmy from making a parody of Coca-Cola, making it where AI cannot generated copyrighted works is insane imo. Those brands should be proud to be such a big part of the cultural fabric that it is difficult…

> Copyright should be the problem of the person using the works and not the problem of the AI generating it.

All parties are responsible on some level. This just reads like passing the buck/burying ones head in the sand to me.

Re: Things are about to get worse for generative AI

#418

Post by RandomLensman »

stale2002 wrote: post #379
schneems wrote: post #370

Earlier quoted context omitted.

Rounding up a transaction and taking the leftovers wouldn’t be a crime worthy of the FBI for one transaction but it would be for a million or a billion. Scale matters and impact matters. If you’re making an ethical argument “it’s okay because it’s already happening to a lesser degree somewhere else” isn’t the flex you think it is. If you’re talking ethics, talk about impact. Who does it help the most who does it hurt…

> “it’s okay because it’s already happening to a lesser degree somewhere else” isn’t the flex you think it is. It actually is. It shows that we as a society are completely OK with this, and nobody is complaining about a very standard and common thing that all artists do. It shows that the outrage is fake, and people don't actually care about the issue.

It shows that as a society we might be ok with humans doing it. Whether or not we are with machines doing it is a different question.

Re: Things are about to get worse for generative AI

#419

Post by stale2002 »

AlienRobot wrote: post #163
4bpp wrote: post #138

Earlier quoted context omitted.

What exactly am I stealing if I don't take the deal, walk away and then enjoy an AI-generated artwork that just so happens to resemble the thing closely instead? I'd think that stealing requires taking something away from someone, regardless of how hard certain industries try to gaslight me into expanding the definition to protect their business model.

Stop trying to gaslight yourself into thinking what you are doing isn't morally wrong. If you do not agree with their business model, don't get involved with their business, at all. Your disagreement doesn't give you the right to exploit flaws in their methods to protect their business. Just like the fact you don't want to pay for something doesn't grant you the right to exploit the fact that the laws of physics allo…

> Your disagreement doesn't give you the right

If I fully believe in the concept of fair use and transformative content, then yes it absolutely is my right to take advantage of generative AI.

Fair use is a common concept used in all sorts of media.

You don't get to hand wave that away just because generative AI is getting good.

Re: Things are about to get worse for generative AI

#420

Post by Spivak »

j_maffe wrote: post #146
Havoc wrote: post #55

To me that’s the wrong question. Everyone knew it was trained on copyrighted material and capable of eerily similar outputs. But it’s already done. At scale. Large corps committing fully. There is no chance of that toothpaste going back in the tube. It’s a bit like when big tech built on aggressive user data harvesting. Whether it’s right, ethical or even legal is academic at this stage. They just did it - effectivel…

That's a really eloquent way of saying "It's already happening, so give up on it." I'm sure it works out great for taking action and solving problems.

It's already happening and most people like having AI more than the DMCA. Selling people on the idea that ML training is piracy to people who on average pirate content with no moral quandary will go nowhere.
Post reply on HN