When you really think about it, one of the fundamental flaws with AI is that it literally cannot create anything new because it requires training data that already exists. The best it can realistically do is mix and match traits of other animals until it creates something “new.” If Star Wars was made today and entirely with AI it would not have the necessary training data to make anything even remotely close to what we got back in the 70s and would look like… This.
And this is a huge problem with pretty much all “AI” today. We say that they can “learn” and be “trained” but it’s basically just large and incoherent database that recycles the data in a way that makes sense to humans. It’s like if you were tasked with creating concept art for a movie or a game but your boss said you can’t start with a blank image and can only modify existing concept art. We already see copy/paste movies, games, and books all the time and we already don’t like it.
You’re exaggerating. Many things people come up with for movies are based on or similar to some things from older shows. You don’t have to make every single creature 100% unique and original. It would actually be strange if people wouldn’t be able to recognize anything on the screen. Look at Guardians of the Galaxy; you have a human, a green humanoid, a red humanoid, a tree-like humanoid and badger like humanoid. Not really levels of creativity beyond AI… Not to mention that the most popular movies today are all remakes and sequels.
The example from Disney was lazy and stupid, AI sux in general but “not being able to create anything new” is not really the main problem here.
And I’m saying that in the context of creativity and using AI for movies it’s not that important. The short movie from the articles is not bad because AI can’t do any better. It’s bad because whoever used the AI tool did a bad job.
All creative works are built on top of works that came before them. AI is no different.
If you tell AI to generate an image of a fat, blue, lizardperson with a huge dick, wearing a fedora, holding way too many groceries… It’ll do that. And you just used it to make something original that (probably) never existed before.
It will have literally created something new. Thanks to the instructions of a creative human.
Saying AI is incapable of creating something new is like saying that programmers can’t create anything new because they’re just writing instructions for a computer to follow. AI is just the latest technology for doing that.
This line of thinking is a real misunderstanding of creativity. “Humans are only remixing things they’ve learned, so AI is the same”. It’s not the same. If an AI has nothing but images of apples in its training data, it will never ever be able to draw a banana. It seems creative and smart on the surface only because its trained on the (stolen) input of every bit of art, text, and code on the internet. But if humans stop creating new imaginative input, it will stagnate right where it is.
If someone has never seen a banana they wouldn’t be able to draw it either.
Also, AIs aren’t stealing anything. When you steal something you have deprived the original owner of that thing. If anything, AIs are copying things but even that isn’t accurate.
When an image AI is trained, it reads though millions upon millions of images that live on the public Internet and for any given image it will increment a floating point value by like 0.01. That’s it. That’s all they do.
For some reason people have this idea in their heads that every AI-generated image can be traced back to some specific image that it somehow copied exactly then modified slightly and combined together for a final output. That’s not how the tech works at all.
Totally disagree. I’ve seen original sources reproduced that show exactly what an AI copied to make images.
And humans can definitely create things that have never been seen before. An AI could never have invented general relativity having only been trained on Newtonian physics. But Einstein did.
Not exactly. As long as the description is understood by AI, it can be generated and won’t be a mish-mash of other animals.
This just showcases the need for precise, and well understood prompt engineering - aka why you can’t just get rid of your designers and have overpaid managers prompt the AI because those chuckfuckles know fuck-all about the process, or creativity (same reason why software engineers won’t be replaced by managers either).
Now obviously, the term to describe what you want has to exist in common vernacular, and for visual AIs, has to have been referenced in the dataset, in some form or manner, for the AI to be able to output it. And ideally, actual designers, graphic artists in the props and costumes department would be using AI for quick visual prototyping, to confirm details, etc., within minutes of the discussion happening (instead of spending hours or days for initial sketches to be declined, refined, finalised), speeding up the process before the final version is then hand-made. This would be the ideal workflow, but some companies have simply grown too big for their own good - like Disney, the leadership is so detached from the actual workforce delivering their products that meddling middle-managers who only care about immediate quarterly and annual costs and incomes, can easily convince said leaders to “help” by getting rid of the expensive people, “replace” them with AI and save tons of money while “delivering the same quality”. Except they’re now realising that it’s nowhere near the same quality, and their customers will actively reject the companies for doing this - but the middle managers don’t care, they delivered “growth” by saving a fuckton of money by firing people thus not having to pay their salaries, got their big fat bonuses for these actions and have already fucked off to the next company to ruin…
When you really think about it, one of the fundamental flaws with AI is that it literally cannot create anything new because it requires training data that already exists. The best it can realistically do is mix and match traits of other animals until it creates something “new.” If Star Wars was made today and entirely with AI it would not have the necessary training data to make anything even remotely close to what we got back in the 70s and would look like… This.
And this is a huge problem with pretty much all “AI” today. We say that they can “learn” and be “trained” but it’s basically just large and incoherent database that recycles the data in a way that makes sense to humans. It’s like if you were tasked with creating concept art for a movie or a game but your boss said you can’t start with a blank image and can only modify existing concept art. We already see copy/paste movies, games, and books all the time and we already don’t like it.
You’re exaggerating. Many things people come up with for movies are based on or similar to some things from older shows. You don’t have to make every single creature 100% unique and original. It would actually be strange if people wouldn’t be able to recognize anything on the screen. Look at Guardians of the Galaxy; you have a human, a green humanoid, a red humanoid, a tree-like humanoid and badger like humanoid. Not really levels of creativity beyond AI… Not to mention that the most popular movies today are all remakes and sequels.
The example from Disney was lazy and stupid, AI sux in general but “not being able to create anything new” is not really the main problem here.
You’re misunderstanding. Op is not saying humans don’t interpolate. They are saying AI can’t extrapolate.
And I’m saying that in the context of creativity and using AI for movies it’s not that important. The short movie from the articles is not bad because AI can’t do any better. It’s bad because whoever used the AI tool did a bad job.
I think your point is valid, but unrelated to previous responses.
All creative works are built on top of works that came before them. AI is no different.
If you tell AI to generate an image of a fat, blue, lizardperson with a huge dick, wearing a fedora, holding way too many groceries… It’ll do that. And you just used it to make something original that (probably) never existed before.
It will have literally created something new. Thanks to the instructions of a creative human.
Saying AI is incapable of creating something new is like saying that programmers can’t create anything new because they’re just writing instructions for a computer to follow. AI is just the latest technology for doing that.
This line of thinking is a real misunderstanding of creativity. “Humans are only remixing things they’ve learned, so AI is the same”. It’s not the same. If an AI has nothing but images of apples in its training data, it will never ever be able to draw a banana. It seems creative and smart on the surface only because its trained on the (stolen) input of every bit of art, text, and code on the internet. But if humans stop creating new imaginative input, it will stagnate right where it is.
If someone has never seen a banana they wouldn’t be able to draw it either.
Also, AIs aren’t stealing anything. When you steal something you have deprived the original owner of that thing. If anything, AIs are copying things but even that isn’t accurate.
When an image AI is trained, it reads though millions upon millions of images that live on the public Internet and for any given image it will increment a floating point value by like 0.01. That’s it. That’s all they do.
For some reason people have this idea in their heads that every AI-generated image can be traced back to some specific image that it somehow copied exactly then modified slightly and combined together for a final output. That’s not how the tech works at all.
You can steal a car. You can’t steal an image.
Totally disagree. I’ve seen original sources reproduced that show exactly what an AI copied to make images.
And humans can definitely create things that have never been seen before. An AI could never have invented general relativity having only been trained on Newtonian physics. But Einstein did.
Not exactly. As long as the description is understood by AI, it can be generated and won’t be a mish-mash of other animals.
This just showcases the need for precise, and well understood prompt engineering - aka why you can’t just get rid of your designers and have overpaid managers prompt the AI because those chuckfuckles know fuck-all about the process, or creativity (same reason why software engineers won’t be replaced by managers either).
Now obviously, the term to describe what you want has to exist in common vernacular, and for visual AIs, has to have been referenced in the dataset, in some form or manner, for the AI to be able to output it. And ideally, actual designers, graphic artists in the props and costumes department would be using AI for quick visual prototyping, to confirm details, etc., within minutes of the discussion happening (instead of spending hours or days for initial sketches to be declined, refined, finalised), speeding up the process before the final version is then hand-made. This would be the ideal workflow, but some companies have simply grown too big for their own good - like Disney, the leadership is so detached from the actual workforce delivering their products that meddling middle-managers who only care about immediate quarterly and annual costs and incomes, can easily convince said leaders to “help” by getting rid of the expensive people, “replace” them with AI and save tons of money while “delivering the same quality”. Except they’re now realising that it’s nowhere near the same quality, and their customers will actively reject the companies for doing this - but the middle managers don’t care, they delivered “growth” by saving a fuckton of money by firing people thus not having to pay their salaries, got their big fat bonuses for these actions and have already fucked off to the next company to ruin…