Originally published at: Marvel defends using AI art for Secret Invasion's opening titles | Boing Boing
…
Secret Invasion does seem like the perfect place to apply AI art, what with the blurring of reality and all. Still…pay someone a fair wage to make the art.
I understand the “they’re coming for our jobs” reactionism, but this luddite-style response to an ideal use case is tiring.
We need to be forward-thinking and prepare for all of the things AI (or, more specifically, “generative models”) can be used for and how it may affect society, but it’s here. And it’s only going to get more useful. Society will adapt. Getting bent out of shape that AI was used to create unnerving and unnatural images that look enough like something that you can tell what it’s supposed to be but it’s also obvious it’s not… is an ideal application of what current-gen AI can do and also a perfect style for the kind of story the show is trying to tell.
I guess the argument is that the company should have spent more money to have more humans use DIFFERENT computer tools to build something that evokes the same feeling, except by hand?
No artists’ jobs were replaced by incorporating these new tools; instead, they complemented and assisted our creative teams.
In legalese, “we didn’t hire new artists when we actually could have, we just went with the current managers”.
We need to admit that AI is often shit at creating artwork.
There have been many shows in the past few years with incredible looking title sequences created by humans using computer technology. Give the people work, and stop trying to hoard wealth.
For a company that built it’s entire business and reputation on the work of visual artists (most of whom were also underpaid), it’s painful to watch them so enthusiastically incorporate a product which is literally built upon the work of such visual artists but doesn’t even acknowledge or compensate creators of the stolen sources of their new, automated “art form”. If that’s how they intend to operate, I’m not even sure the producers who “created” the product should be paid, or the tech companies who are now selling AI “art” services.
AI artwork is an interesting academic exercise, but it’s the creative equivalent of a singer whose “talent” derives entirely from autotune.
Yes. The problem is that the AI tools use existing art and algorithmically use that art to make “new” art, and the artists of the works being used to train the AI are not being compensated at all. AI as it is today can’t create anything genuinely new and original. It’s extrapolating and interpolating. Is AI here to stay? Probably. That doesn’t mean we shouldn’t try to make it work in a way that fairly compensates the humans whose livelihoods it threatens, especially when it is actually using those humans’ work to make its art.
Marvel should know better than this by now. The last time they put an AI in charge of something it tried to wipe out humanity.
“We would talk to them about ideas and themes and words, and then the LOW-PAID GRAPHIC ARTIST would go off and do something. And then we could change it a little bit by using words, and it would change.”
This is going to be looked back in time as super cheesy and dumb, dating it. Using a new toy because they can, and not because it added anything to the show.
9 posts were split to a new topic: “She looks tired…”
Ahhh. Now, see, that’s the nuance that 90% of complaints about AI don’t include (and I’d assume at least 60% of complainers don’t recognize). The complaint is with the training of the tools, not (necessarily) the use of the tools. That’s a more interesting conversation.
Actual artists also train by consuming lots of art on their own. But I don’t think most people would say that’s an equivalent process.
If this opening animation were created via AI that was itself trained exclusively on images and data owned fully by Marvel / Disney, would that satisfy complaints? Certainly there is more to talk about around how the comic industry has traditionally treated its artists, but that’s a separate issue.
Do you think a chatbot trained on your past posts would make an suitable replacement for your contributions to this (or any) forum? Or do you think that something important would be lost if human originality was replaced by an algorithm that had no meaningful sense of insight, creativity or context?
Because it’s not.
No, it wouldn’t. It would effectively be a pay cut for those artists Disney employs.
All of this is, when you boil it down to its essence, another thing that will redistribute wealth from the bottom to the top. It’s the same as self checkout lines at the grocery store, self ordering kiosks at McDonald’s, and heavy automation in factories. All of this is done to reduce costs by reducing labor. Again…you’re absolutely right that none of this is probably going away. But we must face the reality of what this will mean for us. We cannot just put all these people out of work, shovel the money saved in labor costs to the ownership class, and call it a day. Society will collapse at some point.
this kind of thing is also clearly them learning what the state of the tools are, and how they can use them in the future
even if in this case they had hired the same number of artists for the same amount of time for the same pay, it’s absolutely part of the corporate’s self induced pressure to cut all three of those
Ugh, reminds me of my short stint as a restaurant server and the owner kept piling on responsibilities so he could effectively get a “manager” without having to pay for one.
I look at it that way now. I wonder what this would look like and how much better it would be if human-created. This will not go down as innovative, attractive or appropriate.
I know this is the focus of a lot of criticism in this case because, in general, that’s true (and why it’s appealing to companies), but in fairness to Marvel and the team who did this, they could have used the same number of people, using non-AI tools, to produce an opening sequence that looks just as crappy as this one.
No, seriously this looks bad. I don’t know when they did it, but it aged badly before it was even released. It’s all generic (and outdated) AI-generated blobs that’s beyond a cliche at this point.
It’s not even future tense - we hit that point some months back already. AI tools and imagery has evolved to the point where it can create some subtle, interesting effects, if you know what you’re doing. Either this was created long enough ago that the tools didn’t yet exist, or they didn’t know what they were doing.
This really is the perfect use case, as they’re not actually asking it to create something new, they’re trying to just use it to create trippy morphs between images they already produced - something that could (and normally would) be done through regular non-AI software with little effort, but result in a different visual style, by the same number of workers. The problem is, this looks bad and it’s very obviously AI-generated in style, a style which is highly overexposed and the sort of thing that any amateur could produce with access to the tools. Some “AI” tools in this case could have been used like other software to produce an interesting title sequence that looked cool without being problematic, but it looks like it was made using poor tools by someone who didn’t know what they were doing.
They wanted a certain effect - a morphing of images - that was always going to be computer-generated. In this case they were seemingly going for a specific (“AI”) aesthetic purely for the novelty value - which unfortunately for them came off as amateurish and is already over-exposed, making it dated. (And that date being “2022.”)
No, it wouldn’t. It would effectively be a pay cut for those artists Disney employs.
I see. So this is literally a Luddite argument. “These damn weaving machines are going to destroy my ability to make a living.” There’s truth here (there will be lots of disruption), but also a refusal to adapt to reality (there WILL be lots of disruption).
People will still work with AI to create things. The nature of how AI gets trained is an important extra consideration when it comes to that output, but it doesn’t change the fact that someone is still using a tool. I see a lot of people assuming some executive is just going to pull up a text box and type in “make me a top 40 pop song” twelve times a day and then go home.
Will fewer people be able to get more done with less? Yes. That is what technology does. It also means MORE people will be able to make things because the tools will be more accessible. The Internet destroyed newspapers, but enabled new types of citizen journalists. (And means for scams / disinformation.) Technology disrupts. Society adapts. Laws catch up… eventually.
There’s definite reason to inspect Copyright and attribution when it comes to training and using Generative models. The Writers’ strike is, in part, about wanting tighter rules and policies around the use of AI. That makes sense and is important. What I’m objecting to is the tehnophobic overreactions of “OMG someone used a new tool to do a thing! How dare they! How unethical!” without context around what, specifically, they object about the tool.
I’m seeing people effectively saying you can’t “be an artist” and also “use an AI tool” because AI means you don’t need artists? I guess?
Anyway, I expect people to be mad about anything and everything, and generative AI has hit a breakthrough milestone of relevance, so I guess I should just expect to see this sort of thing.
I mean, if AI were used to write novels – which at this stage cannot come up with stories out of whole cloth, and often produces work that just doesn’t make sense…
There’s a reason why folks have been concerned with drawing a line between what is art and what is not art long before the development of digital computing.
ETA: My prediction is that if studios continue to try to “farm out” essential creative processes to AI, we’re gonna wind up with Idiocracy -style media.