The discourse around Palworld shows how AI is already ruining the internet
4K View2024-01-26
Hey, have you heard about this new little indie game, Palworld? Odds are good you have; it’s been the first viral sensation in gaming this year, quickly selling more than seven million copies in its first week on sale, and that’s without counting however many people are playing it on Xbox Game Pass. Whether you’re one of those many Palworld players or not though, you may have encountered Palworld discourse: Commenters from all over the internet have been leveling serious accusations at developer Pocketpair. According to many angry posts on various social media platforms, the studio behind Palworld is guilty of either stealing assets from Pokémon games or using AI-generated assets or possibly some combination of both.
I’m skeptical of both of these claims for separate reasons. The idea that Pocketpair lifted assets from an official Pokémon game is a dubious one: Nintendo is well-known for its proactive approach to guarding its intellectual property, and is quick to swoop in to shut down fan projects, mods, and videos about fan projects and mods for its major titles, Pokémon included. Nintendo has already moved to shut down a fan-created Pokémon mod for Palworld, and issued a DMCA takedown of YouTuber ToastedShoes’ preview video.
Pocketpair is undoubtedly aware of how well-protected Pokémon is, and also probably understands that if it can be proven in court that they copied assets into Palworld, the studio stands to lose every penny it’s made from those seven million copies sold and more. It would be a catastrophically stupid thing to do, and I don’t think it’s likely that Pocketpair would want to expose themselves to that kind of risk. It’s also highly unlikely that Microsoft would be willing to include the game on Xbox Game Pass if it thought that risk was real.
For its part, The Pokémon Company has issued a statement acknowledging the many complaints it has received about Palworld and said that it plans to investigate “and take appropriate measures” if it finds that Palworld has infringed on its Pokémon intellectual property rights. However, the statement does not come close to saying that there has been any infringement.
The accusation of using AI-generated artwork and assets is shakier still. As Paul Tassi lays out this week at Forbes, the claim seems to be based on a series of tweets that Pocketpair CEO Takuro Mizobe made in 2021 about an experiment that used AI to automatically create Pokémon-type creatures. Then, in 2022, Pocketpair released an Among Us-style social deduction game called AI: Art Impostor in which players make artwork using generative AI and must sniff out the impostor in their midst.
It’s actually making a pretty clever point: In AI: Art Impostor, the “true” artists are made aware of a theme for an upcoming gallery showing, while the impostor is not. Everyone produces an AI-generated piece of art, and the game is to figure out which entry was made without the knowledge of the theme—the point being that AI art is frequently the same kind of nonsensical gobbledygook no matter what prompts it’s using.
Instead of seeing this for the satire it is, online Palworld critics have mistakenly taken it as ironclad proof that Palworld uses AI-generated assets. As far as anyone outside of Pocketpair knows, it doesn’t—but once the accusation is out there, there’s no straightforward way to prove that it’s false.
Just to be perfectly clear, I’m not arguing that the accusation isn’t true. For all I know, Palworld could use AI-generated assets somewhere! I don’t have the capability of going through all of the artwork created for the game and evaluating each piece for authenticity. That’ll never be something I can do for any game!
I am, however, deeply skeptical of the accusation. It’s a case of motivated reasoning, a type of bias we’re all prone to engage in from time to time. Motivated reasoning makes us more willing to believe statements that support ideas we already believe or that we want to believe are true; conversely, it makes us more willing to discount any information that contradicts our preexisting beliefs.
The owner of one of the social media accounts that spread some of the accusations against Palworld even admitted that they were doing it because they felt Palworld was “glorif[ying] animal abuse.” That’s as classic a case of motivated reasoning as I’ve ever seen. And once again, it regrettably misses the satirical point Palworld is making about Pokémon and other monster-collecting games.
That’s a topic for another discussion. For now, let’s look at what the presence of generative AI has done to online discussions of games. Palworld is a useful illustration of the phenomenon: It’s very easy to accuse any developer of using AI-generated assets, and the accusation is believable given that we’ve all watched as generative AI has fascinated the likes of Microsoft, Square-Enix, Ubisoft, and countless other developers and publishers.
The uncertainty is a massive problem, and it’s one that I think we’re going to be reckoning with for a long time. The very existence of generative AI makes everything you see or read or play less reliable. If someone says they heard an upcoming game is using AI-generated art assets or voice-over lines, on what basis can you argue that they’re wrong?
I recently did a quick review of Artifact Seeker: Prologue, and there’s a thread in the Steam forums that’s critical of what it calls the “AI art” used in the game. I have no idea what the basis is for the claim that Artifact Seeker uses AI art; the thread’s original poster treats it as established fact without offering any evidence. What’s the solution? Do I devote additional time to tracking down the basis for the accusation? Is it something I should mention in the review?
This is true for the internet more broadly as well. Now that we know that it’s possible to train large language models like ChatGPT to write news stories, every news story we encounter from now on is damaged, even if only slightly, by the suspicion that it might have been created using AI.
I think most of us recoil from AI-generated art and information because we understand that, at its core, it’s counterfeit communication. You’re not looking at something made by a human being, you’re looking at something made to look like a human being made it. It’s a trick, a form of deceit, and we know to keep that kind of thing at arm’s length at best. Now, if you know that a talented counterfeiter is operating in your area, it stands to reason that you’re going to be more suspicious of every twenty-dollar bill that comes your way—you’ll want to take extra time to inspect it, looking for any indications of forgery and for the markings and features used to verify that a bill is genuine.
It’s the same way when we talk about AI now. The existence of tools like ChatGPT and Stable Diffusion makes it more difficult to communicate with each other, because we need additional steps to establish trust and authenticity. This puts anyone interested in sowing doubt and disinformation at an advantage, because they aren’t concerned about that at all—and generative AI now allows them to produce nonsense at an industrial scale.
To put it more simply: Being honest and genuine is harder now and involves additional steps, while dishonesty and fraud are massively easier thanks to the abundance of tools available for creating counterfeit art and information. It’s a bad situation across the board! AI’s lasting effect, I think, is a net increase in the level of what we might call the internet’s information entropy level. The random gibberish, or noise, is taking up a bigger and bigger share of the information available online, while the share of reliable, useful information is dwindling, making it costlier and less worthwhile to even try to find it.
Regardless of whatever else generative AI might wind up doing, it has already managed to poison the well. The internet, as a source of reliable information, is less dependable and harder to use now, and I don’t know whether there’s a way to fix that now that the genie is out of the bottle.
people need to accept that AI is the future. it will not go away. instead of crying every time it is used we should be just learning how to live with it. I'm not going to freak out at blu-rays because I only want to use my VCR. I'm not saying thats necessarily a good thing, just that it's a sure thing.
I don't agree. I think people pushing AI would love it if we assumed it was inevitable, but it's currently a technology with extremely limited use that's being marketed as a solution to everything, which it isn't. Cutting people out of the creative process so that we can have an unending supply of mediocre media is not a future I'm going to resign myself to accepting.
This article is so good. I'm already tired from the number of "Does this game use AI art or not?" debates that are going to pop up for the foreseeable future
people need to accept that AI is the future. it will not go away. instead of crying every time it is used we should be just learning how to live with it. I'm not going to freak out at blu-rays because I only want to use my VCR. I'm not saying thats necessarily a good thing, just that it's a sure thing.
2024-01-28
I don't agree. I think people pushing AI would love it if we assumed it was inevitable, but it's currently a technology with extremely limited use that's being marketed as a solution to everything, which it isn't. Cutting people out of the creative process so that we can have an unending supply of mediocre media is not a future I'm going to resign myself to accepting.
2024-01-29
This article is so good. I'm already tired from the number of "Does this game use AI art or not?" debates that are going to pop up for the foreseeable future
2024-01-27
Author likedyo what
2024-12-21