Live data from Hacker News

Things are about to get worse for generative AI

garymarcus.substack.com

511–520 of 769 posts

Re: Things are about to get worse for generative AI

#511
While different, I find this discussion about AI and copyrights as an evolution of the war that never was: Google/FB converting in the portal/proxy for content and while it is not generative AI you can find copyrighted images just using Google Images or as an snippet in the normal search engine. I mention Google because it is the de facto monopoly but this applies to a lot of aggregators.

I know we are talking about different technologies but it seems all these people were very silent and find some opportunity in having this war with OpenAI (not an endorsement) but not fighting others.

I am not making an statement about the morals of AI and aggregators/search engines (super interesting discussion that in a way was happening for long) but I am surprised that organizations are "just" waking up. It seems they just see it is a much simple and cheap fight.

Re: Things are about to get worse for generative AI

#513

Earlier quoted context omitted.

> They aren't encouraging people to misuse them and it is solely on the user's shoulders for their choice to use them in a way that would cause infringement if the result is used commercially. I agree in principle, but that they can in the first place, especially when it accidentally happens, and at such massive scales more importantly, is the issue methinks. And no one's talking about abolishing the AIs here, we're…

You won't get what you want with those sorts of deals. OK, say every artist gets $100, one time (exact amount varies but would not be much). Everything's properly licensed according to you and the artists are essentially no better off, and the models are now good enough to create new training data for the future and artists never see any more money. You've won, I guess?

Training AI on AI generated data doesn't add anything. The AI already has all the weights to generate the image, so you are at best just reinforcing the existing weights by weighing them more than others.

The closest thing you could do is e.g. have a second model that does something novel like create a 3D model from a 2D image and then you try to animate the model and a third model verifies the quality of the output. This then allows you to selectively reinforce the 2D model using information from the 3D model but this isn't simply generating more training data.

I honestly can't follow your argument. Doing something silly doesn't make you the underdog.

Re: Things are about to get worse for generative AI

#514

These don't seem all that difficult to fix to me. Most of the examples are not really generic, but are shorthand descriptions of well-known entities. "Video game plumber" is practically synonymous with "Mario" and anyone that has the slightest familiarity with the character knows this. Likewise, how difficult is it to just use descriptive tools to describe Mario-like images [1] and then remove these results from anyo…

Seems insane to try to prevent the model from reproducing content with a blocklist like this - to say the least, it’s more than just Mario. Plus, how would you possibly code common-sense fair use into the model? What’s the difference between a cartoon mouse and Mickey Mouse? What if it’s parody? This seems beyond ridiculous to try to enforce on the tool level.

Re: Things are about to get worse for generative AI

#515

Earlier quoted context omitted.

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.

Fortunately, a machine isn't an autonomous mind that does anything on its own. Instead, it is a person who uses the machine, just like fan artists can use a computer to make fan art.

Is that what you’d tell the police if you were caught selling pirated copies of Blu Rays?

Re: Things are about to get worse for generative AI

#516
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.

> Or... things are about to get worse for copyright holders.

If that's so, things are about to get worse for everyone, too. With little to no protection against AI, no one will be incentivized to create new IPs, whether they're books, drawings, songs. Or even films and games, when AI is able to also generate those in the (possibly near) future.

Re: Things are about to get worse for generative AI

#517

Earlier quoted context omitted.

Or they can be forced to destroy or retrain their models without any copyright materials for which they don't have or do not now attain licenses for. These are multi-billion/trillion dollar companies. They can afford to be responsible members of society here, however much their shareholders and C-suite might hate it.

Those weights are never coming out of the BitTorrent network though.

This. The models are out there. Maybe they will just be illicitly shared but even if no new models are trained from scratch I suspect there will be many ways to use extend existing models without going back to scraped images.

I always felt that we already had a solution - I can already get all those images from a web search. Where the law currently intervenes is when I try and distribute works based on close copies of them. Why is this insufficient?

Re: Things are about to get worse for generative AI

#518
post #511

While different, I find this discussion about AI and copyrights as an evolution of the war that never was: Google/FB converting in the portal/proxy for content and while it is not generative AI you can find copyrighted images just using Google Images or as an snippet in the normal search engine. I mention Google because it is the de facto monopoly but this applies to a lot of aggregators. I know we are talking about…

The thing with Google is it is super trivial to exclude your text - tag on page, header on server, etc.. So all the conversations about google "stealing" context always seemed pretty silly to me.

Compared to that AI offers no way to opt out, which is a big difference.

Re: Things are about to get worse for generative AI

#519

Earlier quoted context omitted.

Copyright shouldn't be a problem for anyone. It's simply a bogus idea that artificial scarcity of information is a benefit to society.

Is a man not entitled to the sweat of his brow? 'No!' says the man in Washington, 'It belongs to the poor.' 'No!' says the man in the Vatican, 'It belongs to God.' 'No!' says the man in Moscow, 'It belongs to everyone.' Are you trying to say that no one is entitled to their own inventions? Cause that is a rapid descent into a capitalist hellhole where only those who can steal ideas the most effectively are able to pr…

> Are you trying to say that no one is entitled to their own inventions?

The subject of this thread is copyright, not patents. Though I do believe all intellectual property is bogus (including trademark, which repealing would limit the influence that would be required for the capitalist hellhole you mention), I feel the most strongly so about copyright, which has nothing to do with inventions.

Re: Things are about to get worse for generative AI

#520
post #511

While different, I find this discussion about AI and copyrights as an evolution of the war that never was: Google/FB converting in the portal/proxy for content and while it is not generative AI you can find copyrighted images just using Google Images or as an snippet in the normal search engine. I mention Google because it is the de facto monopoly but this applies to a lot of aggregators. I know we are talking about…

Personal use of copywritten material is fine - there is no breach of copyright when you download a picture from Google for yourself.

If you use it commercially then there is breach.

Uploading copywritten content is a breach of copyright as well, even without commercial use.

Google/Facebook are hosting and giving access to a bunch of media, which might or might not be copywritten - it's the individuals problem.They make.money from ads, not from the content.

AI companies stole copywritten media to train their commercial LLM, sell them or their products and make profit.

I don't think it's the same.

Post reply on HN