Players have been asking for the ability to filter out games made with Gen AI.
We've added an automatic tag on SteamDB based on the AI gen content disclosures on the store pages.
Writing good music is really freakin’ hard, but I do it on my own anyway because the whole point of making something creative is that a person is doing it. It’s truly a shame to see people rally for software made by tech bros that takes work away from real artists who could use it.
editing to be less snarky: How would you feel if generative AI could make a game and an artist or musician had it make an entire game for their art/music because it saved them money?
To a large extent that ship has long sailed in the programming world with Open Source and I even vaguely remember from back in the 90s some people claiming that Open Source would cause programmers to lose their jobs (it didn’t - software users just started to expect even more complex programs with more features and ultimately that resulted in even more programmers being necessary than before), which is eerily similar to the arguments many are making here about AI Gen.
Basically, most of the code in everyday software is already out there and freely available to all in the form of Open Source libraries (which in most projects add up to most of the code in the final executable) and there are even code generators for a number of things, since AI Gen isn’t needed for generating code (because code is a totally artificial thing not something that has to be designed so that the human perception sees it as real or appealing and in fact AI Gen is actually worse at code generation than procedural algorithms) so one can just craft normal code that generates code.
In coding the requirement for using humans has mostly moved from the making of the base parts in a program into the figuring out of how to put the freely available parts together to make a desired greater whole, tough granted the art creation part in game making (some of which I do, since I had to learn 3D modelling for my project and spend a lot of time in it, and the same for Graphical Design which I do for things like icons and UI elements) seems to still rely on a lot of grunt work in low-level shitty shit (and, curiously, the artists in the bigger game-companies are now using expensive AI tools to speed that up).
Let me turn the tables around too: would it be fair if artists and musicians weren’t allowed to use any software which is in full, contains or relies on Open Source code (for example, in the form of libraries), basically the tech level of the 1980s and earlier since almost every software now relies on Open Source code in some way?
Even better, would it be fair for artists who are trying to make it on their own and aren’t superstars?
“By using software which has not been lovingly crafted as whole by a programmer, you’re taking jobs away from programmers.”
(PS: I don’t really want that limitation for anybody)
That said, as I wrote elsewhere, just like programmers are empowered to chose what can be done with the code they make free for everybody as Open Source by choosing the License they ship with it (so, for example, if a programmer wants to force people who make software that contains some of their Open Source code to also release that new software as Open Source, they chose the GPL license, but if they want to give others more freedom to do what they want with it except just sell that freely available code as if it was theirs, the programmer chooses a different license such as the LGPL), so should artists be fully empowered to decide if what they put out there available for all can be used or not in training Generative AI and if they allow it also restrict it to only Generative AI with certain kinds of licensing (say, not for profit, or whose output carries a license that forbids commercial use).
Whilst I would like to use Gen AI for some things in my project, I don’t want to be even indirectly using the works of artists who do not want their stuff used to train Gen AI whose output can be used comercially in any way (so, even as a small part of a greater work).
I don’t want to directly or indirectly take the work of others, I only want to use directly or indirectly the work of willing artists and if there is none, then, well, though luck for me.
In the ideal I would be able to use artwork derived only from the art of artists who would be ok with me using it so, same as you can only use Open Source code (including the tiniest most obscure piece of a library) in the way the programmers are willing for you to use it (so, for example, I cannot distribute commercially a program containing Open Source code - no matter how small - which has been made freely available by the creator under a GPL license, but I can if the license was the Apache one).
Open source software isn’t the same as generative AI at all, but I wouldn’t be surprised if people who put stuff on GitHub pre-ChatGPT don’t want their code scraped for Microsoft to profit off of. I took my personal projects off of GitHub a year or two ago for that reason. Aside from that, open source code is put out there with the intention of being used freely and a lot of art isn’t. Art floating around the internet pre-2022 was made either with the intention of not getting paid for it, in which case whatever, or the artist wanted to get paid and was presumably paid but their art ended up on the internet anyway (but they still got paid at least).
My main issues with generative AI models are that most if not all of them were created without the artists’ or developers’ permission, the profits are going entirely to oligarchs, and training the models takes an obscene amount of energy that contributes to global warming. These models devalue actual art even further and have made the internet a worse place by making it easier to make spam and disinformation. It’s too late to fix spam and disinformation, but we can still value art. I think indie game devs would be singing a different tune if Steam was flooded with games made entirely with AI slop.
edit: For what it’s worth, I paid for my copy of Reaper and the drum plugin I use. I also paid for all my guitar gear.
If you’re mirroring the example being discussed above, then wouldn’t the alternative be that the game doesnt exist in the first place? The musician or artist cant afford to hire a game dev for x amount of time to make a game at all, thats why they used the tool. But using the tool allowed them to get closer to their vision anyway, even if it is imperfect.
I admit that it’s not the best comparison because games often take more time to make than music or visual art. But I think the results are similar if you take a game with programmer art (which does have its charms!) or no art vs one that hired an artist for some amount of time, compared to a soundtrack or set of art with a game the artist put together vs hiring a dev to put together a demo in the same amount of time.
What this really comes down to for me is that most if not all of these models were created without the artists’ permission and training the models takes an obscene amount of energy that contributes to global warming. These models devalue actual art even further and have made the internet a worse place by making it easier to make spam and disinformation. It’s too late to fix spam and disinformation, but we can still value art. I think game devs would be singing a different tune if Steam was flooded with games made entirely with AI slop.
Writing good music is really freakin’ hard, but I do it on my own anyway because the whole point of making something creative is that a person is doing it. It’s truly a shame to see people rally for software made by tech bros that takes work away from real artists who could use it.editing to be less snarky: How would you feel if generative AI could make a game and an artist or musician had it make an entire game for their art/music because it saved them money?
To a large extent that ship has long sailed in the programming world with Open Source and I even vaguely remember from back in the 90s some people claiming that Open Source would cause programmers to lose their jobs (it didn’t - software users just started to expect even more complex programs with more features and ultimately that resulted in even more programmers being necessary than before), which is eerily similar to the arguments many are making here about AI Gen.
Basically, most of the code in everyday software is already out there and freely available to all in the form of Open Source libraries (which in most projects add up to most of the code in the final executable) and there are even code generators for a number of things, since AI Gen isn’t needed for generating code (because code is a totally artificial thing not something that has to be designed so that the human perception sees it as real or appealing and in fact AI Gen is actually worse at code generation than procedural algorithms) so one can just craft normal code that generates code.
In coding the requirement for using humans has mostly moved from the making of the base parts in a program into the figuring out of how to put the freely available parts together to make a desired greater whole, tough granted the art creation part in game making (some of which I do, since I had to learn 3D modelling for my project and spend a lot of time in it, and the same for Graphical Design which I do for things like icons and UI elements) seems to still rely on a lot of grunt work in low-level shitty shit (and, curiously, the artists in the bigger game-companies are now using expensive AI tools to speed that up).
Let me turn the tables around too: would it be fair if artists and musicians weren’t allowed to use any software which is in full, contains or relies on Open Source code (for example, in the form of libraries), basically the tech level of the 1980s and earlier since almost every software now relies on Open Source code in some way?
Even better, would it be fair for artists who are trying to make it on their own and aren’t superstars?
“By using software which has not been lovingly crafted as whole by a programmer, you’re taking jobs away from programmers.”
(PS: I don’t really want that limitation for anybody)
That said, as I wrote elsewhere, just like programmers are empowered to chose what can be done with the code they make free for everybody as Open Source by choosing the License they ship with it (so, for example, if a programmer wants to force people who make software that contains some of their Open Source code to also release that new software as Open Source, they chose the GPL license, but if they want to give others more freedom to do what they want with it except just sell that freely available code as if it was theirs, the programmer chooses a different license such as the LGPL), so should artists be fully empowered to decide if what they put out there available for all can be used or not in training Generative AI and if they allow it also restrict it to only Generative AI with certain kinds of licensing (say, not for profit, or whose output carries a license that forbids commercial use).
Whilst I would like to use Gen AI for some things in my project, I don’t want to be even indirectly using the works of artists who do not want their stuff used to train Gen AI whose output can be used comercially in any way (so, even as a small part of a greater work).
I don’t want to directly or indirectly take the work of others, I only want to use directly or indirectly the work of willing artists and if there is none, then, well, though luck for me.
In the ideal I would be able to use artwork derived only from the art of artists who would be ok with me using it so, same as you can only use Open Source code (including the tiniest most obscure piece of a library) in the way the programmers are willing for you to use it (so, for example, I cannot distribute commercially a program containing Open Source code - no matter how small - which has been made freely available by the creator under a GPL license, but I can if the license was the Apache one).
Open source software isn’t the same as generative AI at all, but I wouldn’t be surprised if people who put stuff on GitHub pre-ChatGPT don’t want their code scraped for Microsoft to profit off of. I took my personal projects off of GitHub a year or two ago for that reason. Aside from that, open source code is put out there with the intention of being used freely and a lot of art isn’t. Art floating around the internet pre-2022 was made either with the intention of not getting paid for it, in which case whatever, or the artist wanted to get paid and was presumably paid but their art ended up on the internet anyway (but they still got paid at least).
My main issues with generative AI models are that most if not all of them were created without the artists’ or developers’ permission, the profits are going entirely to oligarchs, and training the models takes an obscene amount of energy that contributes to global warming. These models devalue actual art even further and have made the internet a worse place by making it easier to make spam and disinformation. It’s too late to fix spam and disinformation, but we can still value art. I think indie game devs would be singing a different tune if Steam was flooded with games made entirely with AI slop.
edit: For what it’s worth, I paid for my copy of Reaper and the drum plugin I use. I also paid for all my guitar gear.
If you’re mirroring the example being discussed above, then wouldn’t the alternative be that the game doesnt exist in the first place? The musician or artist cant afford to hire a game dev for x amount of time to make a game at all, thats why they used the tool. But using the tool allowed them to get closer to their vision anyway, even if it is imperfect.
I admit that it’s not the best comparison because games often take more time to make than music or visual art. But I think the results are similar if you take a game with programmer art (which does have its charms!) or no art vs one that hired an artist for some amount of time, compared to a soundtrack or set of art with a game the artist put together vs hiring a dev to put together a demo in the same amount of time.
What this really comes down to for me is that most if not all of these models were created without the artists’ permission and training the models takes an obscene amount of energy that contributes to global warming. These models devalue actual art even further and have made the internet a worse place by making it easier to make spam and disinformation. It’s too late to fix spam and disinformation, but we can still value art. I think game devs would be singing a different tune if Steam was flooded with games made entirely with AI slop.