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

Database of artists used to train Midjourney AI garners criticism

artnews.com

Post reply on HN
Print view View original

41–50 of 144 posts

Re: Database of artists used to train Midjourney AI garners criticism

#41

Post by galdosdi »

makeitdouble wrote: post #11
thurn wrote: post #3

One positive aspect of the status quo in the United States is that AI-generated images are not currently eligible for copyright. I think this is a great direction to go in, I highly doubt Wizards of the Coast or whoever is going to want their premium products to lose copyright protections, so they'll need to keep paying artists. I'd love for us to lean into this -- you can make all the AI art you want, but it automat…

Won't they move to trademark protection instead, as it's a lot more flexible with less restrictions ? Basically the same way Disney let copyright go but will fight for trademark to the bitter end ?

That won't work. Trademarks are a lot harder to establish -- it's not just automatic from the moment you publish it as with copyright. You have to first start using it, then always mention it's a trademark when using it (with the TM or R symbols for example), then wait for it to catch on, then file with the government some paperwork. (Iirc, exact process is probably similar but different, but the point is it's a lot more involved)

Trademark is intended to protect the holder from being impersonated, not from losing revenue from selling content.... So it's a lot easier to redistribute copies of trademarked work as long as you make it clear you are not affiliated with the trademark holder, in a manner which a reasonable person would heed.

So for example, if a piece of art is trademarked by Disney, and it is well known by the public, and I print a copy and put it in front of my shop, a reasonable observer might this my shop is owned, operated, or endorsed by Fisney. So that's not OK.

If instead I sell copies of that art in my shop, and make it clear to everyone I sell it to that I am in no way affiliated with Disney and this is totally unauthorized by Disney, I'm probably fine.

Trademarks are also industry specific. That's why Apple Records and Apple Computer both exist -- as long as a reasonable person could not confuse them, it's OK.

In short, trademarks are very very different from copyrights. They protect different activities.

In fact I should not have used the phrase trademarked work. A work (like an image or movie or novel or software program) does not get trademarked. The character, slogan, logo, product name, company name, brand name, color scheme, etc used therein to identify the brand, is what is trademarked. Very different.

I will add more examples, this time to illustrate copyright, which works basically the opposite : Suppose mickey mouse were not trademarked. Then while it would be illegal to redistribute verbatim copies of a recent Mickey mouse picture authored by Disney, as well as any modified remixed versions based on that verbatim picture, it would be perfectly legal to draw totally new art involving the same character as long as it was completely new without referring to the copyrighted work, because coypright protects the right of Disney to make money off distributing that picture they made, and they did not make or contribute to making your mickey drawing, and while you are using a character they came up with, in the absence of trademark, copyright isn't intended to protect the public from being confused as to who they are dealing with as trademark is.

IANAL this is based on decades of amateur interest in IP law.

Re: Database of artists used to train Midjourney AI garners criticism

#42

Post by dorkwood »

newZWhoDis wrote: post #18
skydhash wrote: post #14

Earlier quoted context omitted.

Lockpicking tools are widely available, and designs of locks can be found on the web if someone search hard enough. But entering another's home is illegal. It's why businesses pay for licenses even if cracked softwares exists. I hope that artists win these and make using or creating an illegal model an high-risk activity, not worth it for any commercial activity.

[flagged]

If anything, the AI image-generation movement proves just how highly society values the work of the artist, since we now have tech companies dumping hundreds of millions of dollars into training models to approximate their work.

Re: Database of artists used to train Midjourney AI garners criticism

#43

Post by wds »

newZWhoDis wrote: post #18
skydhash wrote: post #14

Earlier quoted context omitted.

Lockpicking tools are widely available, and designs of locks can be found on the web if someone search hard enough. But entering another's home is illegal. It's why businesses pay for licenses even if cracked softwares exists. I hope that artists win these and make using or creating an illegal model an high-risk activity, not worth it for any commercial activity.

[flagged]

As opposed to real jobs like looking at some resumes emailed to you and clicking yes or no on your MacBook

Re: Database of artists used to train Midjourney AI garners criticism

#44

Post by johnnyanmac »

newZWhoDis wrote: post #18
skydhash wrote: post #14

Earlier quoted context omitted.

Lockpicking tools are widely available, and designs of locks can be found on the web if someone search hard enough. But entering another's home is illegal. It's why businesses pay for licenses even if cracked softwares exists. I hope that artists win these and make using or creating an illegal model an high-risk activity, not worth it for any commercial activity.

[flagged]

Compared to many tech people here who maintain a website page for 6 figure salaries?

few of us are saving the world. And it turns out saving the world doesn't pay well either (just look at teacher salaries. I guess educating the future doesn't benefit society). Always found it such a weird angle to demean artists while literal billon dollar corporations profit off of exploiting them. So it's not like people don't value certain kinds of art.

Re: Database of artists used to train Midjourney AI garners criticism

#45

Post by GaggiX »

thurn wrote: post #3

One positive aspect of the status quo in the United States is that AI-generated images are not currently eligible for copyright. I think this is a great direction to go in, I highly doubt Wizards of the Coast or whoever is going to want their premium products to lose copyright protections, so they'll need to keep paying artists. I'd love for us to lean into this -- you can make all the AI art you want, but it automat…

Does this mean that if, for example, a court rules that I cannot train an image generation model on copyrighted material, I can train it on AI-generated images?

Re: Database of artists used to train Midjourney AI garners criticism

#46

Post by tavavex »

kmeisthax wrote: post #19
tavavex wrote: post #12

Earlier quoted context omitted.

You can publish these models in all jurisdictions, including ones that do care about IP law. There's no rulings saying that models trained on datasets of images are direct derivatives of the original images (in a way that's copyright-violating), let alone whatever is produced using these models. Now, this isn't settled law by any means, but it feels like people give too much credence to all the theft accusations by a…

I used to agree with this, but there's been research coming out of Google that has altered my opinion. Specifically, Google's gotten rather good at making AI spit out unaltered training set data[0]. This is only possible if the AI is remembering large portions of the original trained-on works, which would make the weights infringing. [0] In the most egregious case, they found that just asking ChatGPT to repeat a word…

I'm kind of confused - you've claimed that Google's AI was broken, but cite an anecdote over ChatGPT?

Regardless, even if these cases did happen often enough, it's erroneous to assume that this is something universal (i.e. can be applied to all generative AI models) or intentional. Said models are vastly smaller than the sizes of their training datasets, so it's more or less impossible for all the data to be stored verbatim. Some aspects may appear to look like memorization if the piece of data reappears many times in the dataset - in those cases, the algorithm has a really strong incentive to recite that data. This reduces the effectiveness and is considered an artifact that needs to be corrected, not some underlying idea of AI.

The way these algorithms work is public knowledge, there's not really any black boxes in the hands of OpenAI that would be relevant here. Considering that, I'm really doubtful that these claims can be easily supported.

Re: Database of artists used to train Midjourney AI garners criticism

#47

Post by AndyNemmity »

542458 wrote: post #7
thurn wrote: post #3

One positive aspect of the status quo in the United States is that AI-generated images are not currently eligible for copyright. I think this is a great direction to go in, I highly doubt Wizards of the Coast or whoever is going to want their premium products to lose copyright protections, so they'll need to keep paying artists. I'd love for us to lean into this -- you can make all the AI art you want, but it automat…

I think (hypothetically, they may choose not to do this for a number of reasons) WoTC could still dramatically cut back on how much they pay artists by “outsourcing” things like backgrounds, extended art, etc to AI so long as the focal point of the piece is human-created and therefore copyrightable.

Ironically, there's a massive scandal with WoTC right now for doing this. They say they aren't using AI, the majority doesn't believe them due to artifacts only AI would produce for backgrounds.

Re: Database of artists used to train Midjourney AI garners criticism

#48

Post by johnnyanmac »

chii wrote: post #28
kmeisthax wrote: post #19

Earlier quoted context omitted.

I used to agree with this, but there's been research coming out of Google that has altered my opinion. Specifically, Google's gotten rather good at making AI spit out unaltered training set data[0]. This is only possible if the AI is remembering large portions of the original trained-on works, which would make the weights infringing. [0] In the most egregious case, they found that just asking ChatGPT to repeat a word…

> This is only possible if the AI is remembering large portions of the original trained-on works which is fine in my books. You could equivalently produce the same works from searching thru the digits of pi. The only enforcement that's needed is on the end user of the model - if they choose to produce the training set, they are violating copyright. The creator of the model _does not_ violate copyright merely by creat…

>You could equivalently produce the same works from searching thru the digits of pi.

in the same way you can brute force an MD5 Hash if you had a few centuries, I guess. I don't think "monkey's making Shakespeare" is a good metric for when to determine if a piece of art is unique enough to be scraped.

Tracing is looked down upon in the art community and this feels way too close to tracing if they just have these images in a database ready to reference. You can't store copyrighted movies nor music on such private databases (one of the few times I will ever utter the words "thank you DMCA"), I don't see why art pieces would be exempt.

Re: Database of artists used to train Midjourney AI garners criticism

#49

Post by bjt »

chii wrote: post #28
kmeisthax wrote: post #19

Earlier quoted context omitted.

I used to agree with this, but there's been research coming out of Google that has altered my opinion. Specifically, Google's gotten rather good at making AI spit out unaltered training set data[0]. This is only possible if the AI is remembering large portions of the original trained-on works, which would make the weights infringing. [0] In the most egregious case, they found that just asking ChatGPT to repeat a word…

> This is only possible if the AI is remembering large portions of the original trained-on works which is fine in my books. You could equivalently produce the same works from searching thru the digits of pi. The only enforcement that's needed is on the end user of the model - if they choose to produce the training set, they are violating copyright. The creator of the model _does not_ violate copyright merely by creat…

That's a really weird argument. Copyright is a legal system created by the Constitution and statutes and administrative rules. It cares about whether you are "copying", and it cares about whether you're creating things that compete with the works of the original authors. It doesn't care about potential output spaces.

In this context, I don't see a principled difference between the model weights and really good compression. If I send you a gzipped copy of the latest bestseller book it's still copyright infringement. And it would still be infringement if I shipped it inside a software program that can _also_ reshuffle the words in a bajillion different ways, if there's a "copy" of the original work in there.

Re: Database of artists used to train Midjourney AI garners criticism

#50

Post by nwallin »

thurn wrote: post #3

One positive aspect of the status quo in the United States is that AI-generated images are not currently eligible for copyright. I think this is a great direction to go in, I highly doubt Wizards of the Coast or whoever is going to want their premium products to lose copyright protections, so they'll need to keep paying artists. I'd love for us to lean into this -- you can make all the AI art you want, but it automat…

> I highly doubt Wizards of the Coast or whoever is going to want their premium products to lose copyright protections, so they'll need to keep paying artists.

WotC's latest round of layoffs (within the past month or so) hit the art staff especially hard.

Post reply on HN