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

An image of an archeologist adventurer who wears a hat and uses a bullwhip

theaiunderwriter.substack.com

Post reply on HN
Print view View original

411–420 of 927 posts

Re: An image of an archeologist adventurer who wears a hat and uses a bullwhip

#411

Post by bbor »

flessner wrote: post #384

Everyone is talking about theft - I get it, but there's a more subtler point being made here. Current generation of AI models can't think of anything truly new. Everything is simply a blend of prior work. I am not saying that this doesn't have economic value, but it means these AI models are closer to lossy compression algorithms than they are to AGI. The following quote by Sam Altman from about 5 years ago is intere…

Disregarding the (common!) assumption that AGI will consist of one monolithic LLM instead of dozens of specialized ones, I think your comment fails to invoke an accurate, consistent picture of creativity/"truly new" cognition.

To borrow Chomsky's framework: what makes humans unique and special is our ability to produce an infinite range of outputs that nonetheless conform to a set of linguistic rules. When viewed in this light, human creativity necessarily depends on the "linguistic rules" part of that; without a framework of meaning to work within, we would just be generating entropy, not meaningful expressions.

Obviously this applies most directly to external language, but I hope it's clear how it indirectly applies to internal cognition and--as we're discussing here--visual art.

TL;DR: LLMs are definitely creative, otherwise they wouldn't be able to produce semantically-meaningful, context-appropriate language in the first place. For a more empirical argument, just ask yourself how a machine that can generate a poem or illustration depicting [CHARACTER_X] in [PLACE_Y] doing [ACTIVITY_Z] in [STYLE_S] without being creative!

[1] Covered in the famous Chomsky v. Foucault debate, for the curious: https://www.youtube.com/watch?v=3wfNl2L0Gf8

Re: An image of an archeologist adventurer who wears a hat and uses a bullwhip

#412

Post by CaptainFever »

codedokode wrote: post #363
slidehero wrote: post #177

Earlier quoted context omitted.

>There's no reason why training on a billion images is any different You gloss over this as if it's a given. I don't agree. I think you're doing a different thing when you're sampling billions of things equallly.

The root problem is that the model reproduces Indiana Jones instead of creating a new character. This contradicts the statement that the model "learns" and "creates" like a human artist and not merely copies; obviously a human artist would not plagiarize when asked to draw a character.

That does not seem obvious at all. Fan art and referencing is a thing, and there are plenty of examples of AI creating characters that do not exist anywhere in the training dataset.

Re: An image of an archeologist adventurer who wears a hat and uses a bullwhip

#413

Post by awesome_dude »

dcow wrote: post #407
awesome_dude wrote: post #394

Earlier quoted context omitted.

How does someone close source a book?

If the book is the compiled work, then the source of a book is the author's creative process. And certainly that isn't open to all simply by purchasing the book. But less obtusely: you don't copyright a book--which is why knowledge, language, literature should not be closed source. We'd have to find a different model to support authors than trying to prevent people from copying books. Patreon style models where you s…

There are precisely three models for funding

Private - this includes funding by selling item(s), licensing work, and private equity

State

Charity - this includes volunteers, patrons, donations, sponsorships.

Charity relies on people willing to donate for the betterment of others.

State funding fails because of the political nature of the person holding the purse strings.

Licensing, copyright, physical sales are the only thing that artists have to sell.

You "patreon" style falls somewhere between closed source - you can only access if you buy your way behind the curtain, and charity, where creators have to rely on people donating so that their works can be seen by others (for free)

Re: An image of an archeologist adventurer who wears a hat and uses a bullwhip

#414

Post by autoexec »

api wrote: post #346
dcow wrote: post #343

Earlier quoted context omitted.

An artist is a small manufacturing facility that produces a physical (canvas, print, mp3, etc) product, no? What is different about the production of Micky Mouse cartoons? Why is it normal for industries to compete in manufacturing of physical product, but as soon as you can apply copyright, now you exclusively have rights to control anything that produces a similar result?

Let’s say I write a book or record an album and there is no copyright. How do I get paid? Musicians I suppose can tour, which is grueling but it’s something. Authors, programmers, actors, game studios, anything that’s not performed live would immediately become non-viable as a career or a business. Large corporations would make money of course, by offering all you can eat streaming feeds of everything for a monthly f…

>. Let’s say I write a book or record an album and there is no copyright. How do I get paid?

I've purchased books that were in the public domain and without copyright. I've paid for albums I could already legally listen to for free. I've paid for games and movies that were free to play and watch. I'm far from the only person who has or would.

The people who pirate the most are also the ones who spend the most money on the things they pirate. They are hardcore fans. They want official merch and special boxed sets. People want to give the creators of the things they love their money and often feel conflicted about having to give their cash to a far less worthy corporation in the process. There are people who love music but refuse to support the RIAA by buying albums.

There are proven ways to make profit in other ways like "pay what you want" or even "fund in advance" crowdsourced models. If copyright went away or, more ideally, were limited to a much shorter period of time (say 8-10 years) artists would continue to find fans and make money.

Re: An image of an archeologist adventurer who wears a hat and uses a bullwhip

#415

Post by dcow »

jauntywundrkind wrote: post #410
areoform wrote: post #188

Earlier quoted context omitted.

Theft from whom and how? Are you telling me that our culture should be deprived of the idea of Indiana Jones and the feelings that character inspires in all of us forever just because a corporation owns the asset? Indiana Jones is 44 years old. When are we allowed to remix, recreate and expand on this like humanity has done since humans first started sitting down next to a fire and telling stories? edit: this reminds…

Its kind of funny that everyone is harping this way or that way about IP. This is a kind of strange comment for me to read. Because imby tone it sounds like a rebuttal? But by content, it agrees with a core thing I said about myself: > and as someone who thinks IP is itself incredibly ripping off humanity & should be vastly scoped down, it's important to me to not rest my arguments on IP violations What's just such a…

It strikes me that perhaps the prompts are not expansive or expressive enough. If you look at some of the prompts our new wave of prompt artists use to generate images in communities like midjourney, a single sentence doesn't cut it.

If AI is just compression, then decompressing a generic pop-culture-seeking prompt will yield a generic uninspired image.

Re: An image of an archeologist adventurer who wears a hat and uses a bullwhip

#416

Post by godelski »

codedokode wrote: post #261
r3trohack3r wrote: post #246

Earlier quoted context omitted.

You’re setting court precedent that will apply equally to OpenAI as it does to the llama.cpp and stable diffusion models running on your own graphics card.

Can stable diffusion be created without using copyrighted content? Maybe we should have some exemption for non-commercial research but definitely not for commercial exploitation or generating copyrighted images using open-source models.

There is already exemptions for research. Look at licensing around things like ImageNet. There's similar licensing around things like LAION and Common Crawl[0] It's also not legal to just scrape everything without paying. There's a reason the NYT sued OpenAI and then got a settlement. It's still illegal for Meta to torrent terabytes of textbooks too.

[0] https://commoncrawl.org/terms-of-use

  > In this regard, you acknowledge that you may not rely on any Crawled Content created or accumulated by CC.  CC strongly recommends that you obtain the advice of legal counsel before making any use, including commercial use, of the Service and/or the Crawled Content.  BY USING THE CRAWLED CONTENT, YOU AGREE TO RESPECT THE COPYRIGHTS AND OTHER APPLICABLE RIGHTS OF THIRD PARTIES IN AND TO THE MATERIAL CONTAINED THEREIN.

Re: An image of an archeologist adventurer who wears a hat and uses a bullwhip

#418

Post by dvsfish »

Pet_Ant wrote: post #78
jauntywundrkind wrote: post #45

Obviously a horrible hideous theft machine. One thing I would say, it's interesting to consider what would make this not so obviously bad. Like, we could ask AI to assess the physical attributes of the characters it generated. Then ask it to permute some of those attributes. Generate some random tweaks: ok but brawy, short, and a different descent. Do similarly on some clothing colors. Change the game. Hit the "rando…

> Obviously a horrible hideous theft machine. I hate how it is common to advance a position to just state a conclusion as if it were a fact. You keep repeating the same thing over and over until it seems like a concensus has been reached instead of an actual argument reasoned from first principle. This is no theft here. Any copyright would be flimsier than software patents. I love Studio Ghibli (including $500/seat f…

The engagement ring is a good example object, but I feel it serves the opposite argument better.

If engagement rings were as ubiquitous and easy to generate as Ghibli images have become, they would lose their value very quickly -- not even just in the monetary sense, but the sentimental value across the market would crash for this particular trinket. It wouldn't be about picking the right one anymore, it would be finding some other thing that better conveys status or love through scarcity.

If you have a 3d printer you'd know this feeling where abundance diminishes the value of something directly. Any pure plastic items you have are reduced to junk very quickly once you know you can basically have anything on a whim (exceptions for things with utility, however these are still printable). If I could print 30 rings a day, my partner wouldn't want any of them as a show of my undying love. Something more special and rare and thoughtful would have to take its place.

This isn't meant to come across as shallow in any way, its just classic supply and demand relating to non monetary value.

Re: An image of an archeologist adventurer who wears a hat and uses a bullwhip

#419

Post by vkou »

Lerc wrote: post #322
MgB2 wrote: post #30

Idk, the models generating what are basically 1:1 copies of the training data from pretty generic descriptions feels like a severe case of overfitting to me. What use is a generational model that just regurgitates the input? I feel like the less advanced generations, maybe even because of their limitations in terms of size, were better at coming up with something that at least feels new. In the end, other than for co…

I'm not sure if this is a problem with overfitting. I'm ok with the model knowing what Indiana Jones or the Predator looks like with well remembered details, it just seems that it's generating images from that knowledge in cases where that isn't appropriate. I wonder if it's a fine tuning issue where people have overly provided archetypes of the thing that they were training towards. That would be the fastest way for…

> I'm ok with the model knowing what Indiana Jones or the Predator looks like with well remembered details,

ClosedAI doesn't seem to be OK with it, because they are explicitly censoring characters of more popular IPs. Presumably as a fig leaf against accusations of theft.

Re: An image of an archeologist adventurer who wears a hat and uses a bullwhip

#420

Post by sothatsit »

jauntywundrkind wrote: post #410
areoform wrote: post #188

Earlier quoted context omitted.

Theft from whom and how? Are you telling me that our culture should be deprived of the idea of Indiana Jones and the feelings that character inspires in all of us forever just because a corporation owns the asset? Indiana Jones is 44 years old. When are we allowed to remix, recreate and expand on this like humanity has done since humans first started sitting down next to a fire and telling stories? edit: this reminds…

Its kind of funny that everyone is harping this way or that way about IP. This is a kind of strange comment for me to read. Because imby tone it sounds like a rebuttal? But by content, it agrees with a core thing I said about myself: > and as someone who thinks IP is itself incredibly ripping off humanity & should be vastly scoped down, it's important to me to not rest my arguments on IP violations What's just such a…

> All this power! And yet!

You are completely ignoring the fact that you can provide so much more information to the LLMs to get what you want. If you truly want novel images, ChatGPT can absolutely provide them, but you have to provide a better starting point than "An image of an archeologist adventurer who wears a hat and uses a bullwhip".

If you just provide a teensy bit more information, the results dramatically change. Try out "An image of an Indian female archeologist adventurer who wears a hat and uses a bullwhip". Or give it an input image to work with.

From just adding a couple words, ChatGPT produces an entirely new character. It's so easy to get it to produce novel images. It is so easy in fact, that it makes a lot of posts like this one feel like strawmen, intentionally providing so little information to the LLMs that the generic character is the only obvious output that you would expect.

Now, would it be better if it didn't default to these common movie tropes? Sure. But the fact that it can follow these tropes doesn't mean that it cannot also be used to produce entirely new images filled with your imagination as well. You just have to actually ask it for that.

Post reply on HN