Live data from Hacker News

No elephants: Breakthroughs in image generation

oneusefulthing.org

71–80 of 373 posts

Re: No elephants: Breakthroughs in image generation

#71
post #2

That proper "no elephants" first image is hilarious. Another key point here is the generative AI's meme game is getting rather strong. Which isn't a small thing, humour is an advanced soft skill.

The "Game Design Otter" action figure seems to come with a pair of flashlights. I bet that's a residue from the previous prompt about illuminating the tablet with a flashlight.

Re: No elephants: Breakthroughs in image generation

#72
post #34

It's interesting to hear people side with the artists when in previous discussions on this forum I've gotten significant approval/agreement arguing that copyright is far too long. As I've argued in the past, I think copyright should last maybe five years: in this modern era, monetizing your work doesn't (usually) have to take more than a short time. I'd happily concede to some sort of renewal process to extend that p…

It's one thing to argue that copyright terms should be shortened, and another to accept that a handful of corporations should be able to forcefully shorten it for certain actors entirely on their own terms.

Re: No elephants: Breakthroughs in image generation

#73
post #34

It's interesting to hear people side with the artists when in previous discussions on this forum I've gotten significant approval/agreement arguing that copyright is far too long. As I've argued in the past, I think copyright should last maybe five years: in this modern era, monetizing your work doesn't (usually) have to take more than a short time. I'd happily concede to some sort of renewal process to extend that p…

> I think copyright should last maybe five years: in this modern era, monetizing your work doesn't (usually) have to take more than a short time. I funny how people who say this kind of stuff are never content creators (in the monetization sense).

There are a lot of programmers on this platform (myself included), and I love that my work has an impact on others.

I have a number of public repos and I have benefitted greatly from other public repos. I hope LLMs made some use of my code.

I wrote blogs for years without any monetization. I hope my ideas influenced someone and would be happy if they made some impact on the reasoning of LLMs.

I'm aware of patent trolls and know people with personal experience with them.

So I generate a lot more content that the typical person and I am still in favor of much looser IP rights as I think they have gone overboard and the net benefit for me, a content creator, is much greater having access to the work of others and being able to use tools like LLMs trained on their work.

Re: No elephants: Breakthroughs in image generation

#74
post #34

It's interesting to hear people side with the artists when in previous discussions on this forum I've gotten significant approval/agreement arguing that copyright is far too long. As I've argued in the past, I think copyright should last maybe five years: in this modern era, monetizing your work doesn't (usually) have to take more than a short time. I'd happily concede to some sort of renewal process to extend that p…

> I think copyright should last maybe five years: in this modern era, monetizing your work doesn't (usually) have to take more than a short time. I funny how people who say this kind of stuff are never content creators (in the monetization sense).

If you make a blog with nice original long form articles it may take much longer to gain traction. Reproducing the content in "your own" wording quickly gets fuzzy.

I like the practical angle. Any formula that requires monitoring what everyone is doing is unworthy of consideration. Appeal to tradition should not apply.

Re: No elephants: Breakthroughs in image generation

#75

Diagrams are still a big unsolved problem. Making diagrams for a talk or paper is an extremely tedious process and I am still waiting for a good multimodal LLM solution for this. It should take a sketch and/or text description of what you want and in a few iterations you should get what you want. GPT4o tries hard but results are still bad.

I've had the best luck at getting it to produce diagrams-as-code like mermaid or plantuml.

Re: No elephants: Breakthroughs in image generation

#76
post #73

Earlier quoted context omitted.

> I think copyright should last maybe five years: in this modern era, monetizing your work doesn't (usually) have to take more than a short time. I funny how people who say this kind of stuff are never content creators (in the monetization sense).

There are a lot of programmers on this platform (myself included), and I love that my work has an impact on others. I have a number of public repos and I have benefitted greatly from other public repos. I hope LLMs made some use of my code. I wrote blogs for years without any monetization. I hope my ideas influenced someone and would be happy if they made some impact on the reasoning of LLMs. I'm aware of patent trol…

posting stuff for free is different than selling stuff.

Re: No elephants: Breakthroughs in image generation

#77
post #3

This is a before/after moment for image generation. A simple example is the background images on a ton of (mediocre) music youtube channels. They almost all use AI generated images that are full of nonsense the closer you look. Jazz channels will feature coffee shops with garbled text on the menu and furniture blending together. I bet all of that disappears over the next few months. On another note, and perhaps other…

Image generation is still very slow. If it generated many images instantly like Google’s image search, it would be a lot more fun to use, and we would learn to use it more effectively with practice.

Re: No elephants: Breakthroughs in image generation

#78
post #34

It's interesting to hear people side with the artists when in previous discussions on this forum I've gotten significant approval/agreement arguing that copyright is far too long. As I've argued in the past, I think copyright should last maybe five years: in this modern era, monetizing your work doesn't (usually) have to take more than a short time. I'd happily concede to some sort of renewal process to extend that p…

Mm.

The arguments about wanting copyright to be life+70 have always felt entitled, to me. Making claims about things for their kids to inherit, when the median person doesn't have the option to build up much of an inheritance anyway, and 70 years isn't just the next generation but the next 2.5 generations.

I don't know the exact duration of copyright that makes sense, the world changes too much and different media behave differently. Feels like nobody should have the right to block remakes of C64 games on copyright grounds, but I wouldn't necessarily say that about books published in the same year.

From what I've seen about the distribution of specifically book sales, where even the top-100 best sellers often don't make enough to justify the time involved, I think that one of the biggest problems with the economics of the arts is a mixture of (1) the low cost of reproduction, and (2) all the other artists.

For the former: There were political campaigns a century ago, warning about the loss of culture when cinemas shifted from live bands to recorded music[0]; Today, if I were so inclined, I can for a pittance listen to any of (I'm told) 100 million musical performances, watch any of 1.24 million movies or TV shows. Even before GenAI, there was a seemingly endless quantity of graphical art.

For the latter: For every new book by a current living author such as Charlie Stross (who is on here sometimes), my limited time is also spread between that and the huge back-catalogue of old classics like the complete works of Conan Doyle, Larry Niven, or Terry Pratchett.

[0] https://www.smithsonianmag.com/history/musicians-wage-war-ag...

Re: No elephants: Breakthroughs in image generation

#79

> Is it okay to reproduce the hard-won style of other artists using AI? Who owns the resulting art? Who profits from it? Which artists are in the training data for AI, and what is the legal and ethical status of using copyrighted work for training? These were important questions before multimodal AI, but now developing answers to them is increasingly urgent. I have to disagree with the conclusion. This was an importa…

“Fair” doesn’t matter. The only consensus that matters is what is legal and profitable. The former seems to be pretty much decided in favor of AI, with some open question about whether large media companies enjoy protections that smaller artists don’t. (The legal battle when some AI company finally decides to let their model imitate Disney stuff is going to be epic.) Profitable remains to be seen, but doesn’t matter much while investors’ money is so plentiful.

Re: No elephants: Breakthroughs in image generation

#80
post #34

It's interesting to hear people side with the artists when in previous discussions on this forum I've gotten significant approval/agreement arguing that copyright is far too long. As I've argued in the past, I think copyright should last maybe five years: in this modern era, monetizing your work doesn't (usually) have to take more than a short time. I'd happily concede to some sort of renewal process to extend that p…

> I think copyright should last maybe five years: in this modern era, monetizing your work doesn't (usually) have to take more than a short time. I funny how people who say this kind of stuff are never content creators (in the monetization sense).

Not that it should impact the validity of my argument, but I have sold commercial software in the past, and it is absurd that that software will be copyrighted through most of the 21st-century.
Post reply on HN