Added Palworld

Based on this article on Video Games Chronicle, I've added Palworld to this concept.

https://www.videogameschronicle.com/news/palworld-embroiled-in-ai-and-pokemon-plagiarism-controversy/

This is developing and I'll update with any confirmations or any other new articles, but evidence is basically flooding in on social media.

https://twitter.com/DeviYun_Vtuber/status/1748757748370547046

https://twitter.com/GenePark/status/1748858054177620195

its probably just normal ripoffs instead of AI

They definitely have blatantly taken design elements directly from other games, but at this point, it doesn't seem like there's much specific evidence of the use of AI.

Yeah this is why I posted here about it's specific inclusion. I am not super well equipped to point out what is Generative A.I. trained on very specific copyrighted data vs. plain old stealing of assets or at least being "heavily inspired by". I really only have my understanding of Machine Learning (which is only basically what I got from my degree and not working in the field everyday or anything).

So I wanted to post here to kind of see what people thought. I should have been more clear on that.

With going over the evidence and with a day to think on it, I think I may have jumped the gun a bit. So I will be removing it for now.

Still, I will in no way be surprised if it ends up that the game was made using Generative A.I. trained on Pokemon designs. I will keep posting about it here if more evidence or news stories come to light (which probably won't be until the week starts).

I agree that I wouldn't be totally surprised if they used AI, but at this point, it does seem more likely that this is just a classic case of copying someone else's homework.

I've watched a few hours of people playing it in streams and i'm not sure what parts would be using AI generated assets, the world is on par with other Unreal engine asset flips, it does seem to be copying designs from Pokemon though, the example in that article is pretty obvious and i imagine they'll be changing that art soon.

I wasn't expecting it to basically be Ark, i never got that impression from the pre-release footage, depending on how much content is in this early access version of the game i fully expect its popularity to drop pretty drastically, it looks fun but Ark and other similar games have years of continued development over Palworld.

I don't see how/why thats AI or any different than procedural generation in No Man's Sky, people are just waiting to be up in arms over AI.

One of the most expensive parts in making games is creating assests a studio using AI isn't that far fetched particularly a small one.

So far, the evidence for this game containing AI seems circumstantial (the developer's standpoint on and past use of AI being the most often cited, as far as I've seen). I've seen statements from the dev glazing up the "one grad student" that does all their art. The simplest explanation seems to be ordinary human plagiarism, whether accidental or intentional. I don't blame the art student (seems like they're being exploited), but since anyone with a functioning brain immediately picked out the extreme similarities with Pokemon, the developer has to know what they're doing.

Honestly, the most compelling argument for AI in the game is the old series of tweets (which may be poorly translated) talking about how AI could help get around copyright restrictions. I'm astonished Nintendo / Game Freak haven't sued this game, though maybe its suddenly elevated profile will change that. Not like Nintendo has ever held the legal dogs back before, so who knows. (EDIT: VGC has an article about how the in-game assets may have been directly ripped from a Pokemon game. That might change things pretty significantly!)

I suspect we're all going to be very tired of hearing about Palworld by the end of this month.

Destructoid posted a pretty level-headed take on the whole thing.

https://www.destructoid.com/best-places-to-respawn-early-on-in-palworld/

@thepanzini:While the output can seem similar, the difference is in how the sausage is made.

No Man's Sky has a procedural generation system that takes pre-made assets made by a person and stitches them together in new ways. This is true for the plants, animals, terrain, planets, ships, aliens, space stations, etc. etc. All of those assets were made by a person, even if that combination was never what was intended. It's essentially just a LOT of smart coding and efficient use of created assets that is then thrown into a random number generator. The seed of that RNG tells the algorithm what to make where and in what way.

Generative A.I. doesn't work like that. It is an algorithm someone created, yes, but not to specifically to create a No Man's Sky like galaxy. Now, again, I'm no Machine Learning expert, but the best, simple description I can give of what Generative A.I. is this: Generative A.I. is a complex mix of pattern recognition in data (like what objects exist in a picture), natural language interpretation (giving the program plain old English and it converting that into data it can understand and use), and repurposing data it has analyzed and accumulated into an output that fits the parameters of its input.

So... what does that mean? It means that for a Generative A.I. to work, you have to feed a LOT of examples into it, all labelled with different information about what is in the input (ex. this image is of a band playing on a stage). Once the Generative A.I. has enough data points of what a band looks like playing on a stage, it will start being able to repurpose data it has already taken in and spit out a new output on what an image of a band playing on a stage looks like. It's not creating something new, just stitching colour values, patterns, and details together to be a certain way that is in line with what it has been told what a band being on stage looks like.

Now with this, Generative A.I. and Procedural Generation might still sound the same, but there are four crucial changes to how Generative A.I. works.

1) Procedural Generation uses input that is all made by the developer, using exactly the media and assets that were made to work with it.

Generative A.I., however, for it to spit out a "new" image of a band, it has to have a VERY LARGE amount of images that are of bands on stages that have been labelled and processed. Like WAY MORE than you even think. Way more than any one person could take photos of or draw images of in their life. Where do you get those images from then? It has to be sourced from somewhere. And very few people have given permissions for companies to use their work, regardless of the medium, for use to train these neural networks.

So every single A.I. neural network has been trained off of information and media they have either been scraped from sites that don't even know their work is being fed into an A.I. OR it's a company like Google, Microsoft, Apple, or Meta that have large amounts of data on their servers that they can use without informing people (or at least informing them buried deep in a 200 page terms of service document).

2) In Procedural Generation, the output is curated by the developer by honing the algorithms it uses to ensure the level, animal, whatever is how they want the game to be.

The A.I. has no control or filter over what it outputs. It just sees data. Data that is interpreted one way from input, more data from a prompt from a user, and output that fits the data it has. And there are very few nobs that one can turn to filter this.

Let's say you used A.I. to make Spelunky levels based on existing Spelunky levels. No matter how much valid, awesome, procedurally generated Spelunky levels you feed into the algorithm, it is still going to produce some levels with, say, no way to reach the exit. Or a drop that you can't fall down without hurting yourself. Or an enemy that shouldn't exist in one level being in another. Or lava right above the entrance.

Because the A.I. doesn't have rules to make a thing. It just is given data and output based on what data it has been given before. Sure, none of the levels it had been given had a section where it was impossible to get down without dying, but it doesn't look for that. It just looks at the data and rearranges it to look like data it has seen before.

To give the A.I. all of the data it needs to perfectly make Spelunky levels, you have to feed into it, with each of those levels, all of the information that you already have procedural generation checking for. On top of all those completed levels you have already generated using those algorithms. And all of the assets to make them look authentic.

3) Procedural Generation is stable. Meaning that, except for bugs or oversights in its creation, procedurally generation will always produce output that it is expected to make.

Generative A.I. is constantly changing with each piece of input it is fed and there is no way to reign in if those changes are *bad*.

We've seen this already with text-based Generative A.I., over time losing their ability to do basic arithmetic (you know, the easiest thing a computer can do). And we're going to see this with image-based Generative A.I. very soon too, with artists fighting back against it with A.I. poisoning tools like Nightshade for when their art is inevitably stolen from the internet by their scraping algorithms gathering input to feed their neural network.

And while Nightshade is a malicious act against neural networks (but not unwarranted imo), such acts are, again, not necessary for the A.I. to deteriorate. Just give enough humans the ability to chat with it for a long enough time and the work will be done. Because if you were able to make an A.I. that had checks in place that was good enough to distinguish all data as valid or invalid from the human race, you wouldn't need the A.I. part of it. You would just have just manually programmed something so complex that it has omnipotence.

4) Procedural Generation doesn't fucking steal anything.

@devoureroftime:

Any developer can already buy models created using AI with the sourced images used to create them, none of it needs to be stolen source.

Palworld only has ~100 Pals the perk of using AI or procedural generation is you can create a alot of something, and I very much doubt either was used.

Given the game was created by a team ~12 a few being students it'd be much simpler just to copy a few Pokemon and change them slightly which is gaming 101.

If the reference material used is copyright content then procedural generation wouldn't essentially be any different, likewise in reverse for AI.

AI is definitely gonna have a place in gaming, a small indie developer using it to do something it otherwise wouldn't be able to do seems like a valid one.

This is something I don't understand about complaints of AI using copyrighted material to generate assets - humans do the same thing. No artist has gotten their education by only looking at public domain art. Musicians don't list Audio Network or Wikimedia Commons as sources of inspiration during interviews. I don't know why computers getting ideas from existing content is any worse than humans doing it.

It's an extremely complicated topic but a lot of it comes down to licensing, attribution, and the potential for harm to living, working artists. A great example is Getty Images. It was found that one of the largest image generators had scraped Getty's various thousands upon thousands of photos from their website to use as training data without permission from Getty or any acknowledgement that they had done it (this was found out because the image generator was putting in broken, janky looking versions of the Getty watermark on generated images). Getty paid photographers for all of those photos and licensed them such that they should not have been used as training data. So we end up with a case of both Getty having the thing they sell being stolen and the work of thousands of photographers being used in a tool that could put some of these photographers out of work (why pay Getty for generic photos when you can just generate photos that accomplish the same thing for articles). It should be obvious that this is harmful to working photographers. It's no different for any other form of art.

I don't know why computers getting ideas from existing content is any worse than humans doing it.

Humans don't get a free pass on this so why should computers? If an artist blatantly copies the work of another without attribution and gets found out, they have an extremely short window of time to fix the problem before it often becomes a career-ending thing for them. A good recent example of this is Olivia Rodrigo. She had to retroactively give several other artists songwriting credits on her debut album because several of her songs were found to be suspiciously similar to those already existing (the most obvious example being "good 4 u", which took the entire song structure, various chord sequences, and other bits from Paramore's "Misery Business". It was extremely blatant to the point that you could play the two songs side by side and they matched up almost perfectly at times). She said that these artists were huge inspirations for her and was doing homages to them, etc. etc., but the reality was that had she not given them credit she would have been in heaps of trouble and probably had her career ended.

Taking inspiration from and copying are two entirely different things. Current generative AI is doing mostly the latter, not the former. Taking someone else's work and shuffling around things a tiny bit doesn't make what you are doing not plagiarism. For example, if I take the idea of Nile Rodgers-style jangly disco guitar and put it in my own distinct disco song, that's generally fine. If I completely rip off the song "We Are Family" but change the lyrics and shift a couple chords, chances are Nile's not gonna be happy with me unless I get his permission and give him the accreditation that will allow him to be paid for his work that I'm copying. If you asked a generative AI to write a song in the style of Nile Rodgers, it would probably do something similar to the latter example because that's how these models work. They can't create original work because everything they make is inherently derivative.

Remember, when thinking about this stuff you have to give no benefit of the doubt to the people pushing it because they are catering to the shittiest, most cynical business people who would happily fire a bunch of artists to make a number go up slightly. They don't care about art or quality, all they care about is number going up. We've already seen what's happened to creative jobs in the game industry the last year (and this was without generative AI). We don't need to see that happen in every other creative industry too because some suits think they can save a bit of money by using a bunch of shitty generic generative AI assets.

@ben_h: So would it not be an issue if the AI was able to generate things further away from the initial source material? The way I see it is that the training data is the way AI learns what things look like, just like how humans spend years looking at what existing things look like to create their own similar things.

I think part of the problem I have is that I just fundamentally don't understand copyright and attribution and the idea of stealing art. I don't understand why people want credit for their work, just like you probably don't understand why I don't understand that. There's just some step in the logic that I can't follow.

I just don't think AI can really think like us. It's all a little too literal. The image in my head doesn't match up with what AI tends to spit out.

Case in point, I asked craiyon to give me 'Joseph Stalin as a six-headed Hydra.' A very specific request that somehow gave me Joseph Stalin accompanied by a six-headed hydra. At best, it was Joseph Stalin with six hydra heads attached to his normal, Joseph Stalin looking head.

The other 8 options were vanilla Joseph Stalin portraits, all variations on a theme. Not the eldritch horror I was hoping for.

Compare it to something like Jim'll Paint It. Something very much basic around requests but with more more detailed and inspired results.

https://jimllpaintit.tumblr.com/

@eccentrix: Do you understand money and wanting to eat and pay rent?

It takes a lot less effort to copy than to create, so if there is no copyright to protect artists and other creators they can't get paid for their work. If you record an album and someone else can just stream it and collect the revenue they can undercut you (because they don't need to spend the time or money to record it) and make it economically impossible for you to make a living from your recorded music. Just tour? Well if anyone can play your songs (and claim to have written them) that's much harder too. The Rolling Stones can only command the prices they do because everyone knows they're the Rolling Stones, and anyone else who wants to play their music has to pay them (they don't always, but they're supposed to.)

Now the Rolling Stones have enough money, but there are lots of smaller acts and artists who don't and who need that revenue if they are going to make a living as artists.

And even if you're not getting paid there's social status from having created something and recognition that comes with it.

People want to be rewarded for their creative work the same way people want to be rewarded for other work they do.

Imagine if you worked an 8 hour shift at a warehouse and someone came up at the end, took credit for your work, and collected your paycheck. That would not be cool. Or if you want to take money out of it, imagine if you mowed a friend's lawn every week while they were out of the country and then when you came to welcome them back someone else took credit for it. Would that be a fun time?

@bigsocrates: that's not the same at all, and I think you know if isn't

@bigsocrates: It just feels backwards to me, like it's contrived in order to make people money, not because people want to pay for it but because people want to be paid for it.

"I want to be paid for this thing I do." Why? Because you do it? Why are you turning the thing you're doing into something that makes money instead of doing something that makes money already? The system is only built to pay artists because artists want to be paid. I wouldn't train as a lawnmower and then mow lawns on spec expecting to be paid later. It's like any industry, either you can get a paying job doing the thing or you can't. Sometimes whole careers just disappear. Illegal weed dealers are having to move onto other things. Nobody's mourning the loss of the manual typesetting industry anymore, but I'm sure people still do it as a hobby.

That's just the configuration my mind is in. I'm not saying capitalistic job creation for the sake of it is wrong, it's just how my mind naturally approaches the situation and it's hard to consider it differently.

Taking money out of it, I suppose if I was only mowing lawns for the credit, I'd be upset if someone else took that credit, but that's the thing - I don't understand the appeal of credit. I don't understand the social status and recognition that come with it.

@eccentrix: Nobody wants to pay for anything. Everyone would rather get things for free. You can say the exact same thing about any commercial exchange. "I want to be paid for cutting hair." Why? Because it's my skill and I want to be paid for it.

The system is NOT only built to pay artists because they want to be paid. It's built to pay artists in order to encourage art and, more importantly, science and engineering. It's built the way it is because otherwise artists and inventors and everyone else would have to have other jobs to support themselves or find patrons and that would cause society to have a lot less art and innovation.

This is the same reason we have private property. There's no fundamental difference between the courts protecting someone's copyright and the courts protecting your property rights on a philosophical level. They both reflect choices made to organize society in certain ways for certain ends.

And it's very strange to say "why not do something that makes money already" as if art hasn't made money for...centuries. Copyright is in the US constitution. You're acting like it's some newfangled thing. Not all artists do what they do out of love. Some do it precisely because it's a career that makes them money.

Comparing it to illegal weed dealing makes no sense because that was never a job that reflected a choice to organize society in a certain way to promote it. It was the opposite. And comparing it to typesetters makes no sense because people still want to consume art. This is more like saying that people should give up working at banks because they've become easier to rob through electronic means (which they have; fraud and hacking have made things harder for financial institutions.)

Copyright has been part of society for centuries and is a cornerstone of our legal system and how our society is organized culturally. The idea that people should just accept that their long term careers are damaged or over because it's gotten harder to enforce, or that they're weird for following very established career paths and wanting to be paid for their work seems very strange when applied to literally any other profession or occupation.

It doesn't have to be that you were mowing lawns "for the credit." It would still anger most people to see someone else taking that credit even if it wasn't a primary motivation. Why do people want credit for their actions? Because it's part of social status and relationships. It's part of being accurately assessed by others. If you don't understand why people care what others think of them or why people want to be accurately perceived then human social relations must baffle you.

The system isn't new. As you say, it's centuries old. I'm new to the system. I was born relatively recently in terms of human society and I'm coming to it fresh after it's had many centuries of progression. The system isn't strange, because it's presumably developed by some kind of consensus, deliberately or not. The way it's developed isn't intuitive to me. I'm the strange one.

Also, people do still want to consume printed type, it's just made by computers now, so it's perfectly analogous to the original topic.

@eccentrix: But you're asking the question why aren't these people already doing something that people get paid for if their goal is to be paid, and the answer is that they ARE doing something people have been paid for for a very long time. You might as well ask why people want money, and the answer for that is self-evident.

People want printed type but they don't want typeset type. That would be analogous to when animation shifted from hand drawn to mostly computer generated (even animation that looks hand drawn is, these days, at the very least computer enhanced, even leaving aside stuff like Pixar.) Did people lose jobs? Yes. But because they lost jobs to a different more efficient process the complaints weren't the same as when work is flat out stolen.

There are lots of examples in art analogous to typesetters in art including entire art forms that have fallen out of fashion for various reasons (movies destroyed vaudeville) and that's just how things go in society and life. But theft is different because the demand and payment still exists it's just being channeled to people who aren't producing the thing, and because if allowed unchecked it runs the risk of destroying cultural production.

@bigsocrates: I'm saying people don't get paid to be artists because people become artists because it's a job. It had to start somewhere and if it doesn't already exist as a job, it doesn't make sense to take it up as a career. You're saying "But it does exist as a job" and I'm saying "But it didn't always". We're just coming at it from different directions.

Theft might destroy cultural mass production, but it doesn't stop people from making art or other people from enjoying it. People make and consume free art all the time and I'd posit that most art is never seen by anyone but the artist. AI might change some artistic industries, but it won't change humanity.

@eccentrix: Nothing always existed as a job. It's not like there were butchers in the very earliest societies. What happened was that people were better at one thing than other people were and slowly others began to compensate them for doing those things. This is how specialization happened. Artist has been a "job" for a very long time. If you want to go pre-copyright artists were supported by the church and then secular patrons, or they were "artisans" making crafts items to sell (we're talking European history here, it was different in other societies) because people wanted the art. If you want to go so far back that there were no specialized artists making a living off their work then we're going to go to like pre-Roman society and again have to question every profession. I don't understand why this is a specific issue you seem to have with artists and not, say, doctors.

Destroying cultural mass production will uproot a lot of society. It will also destroy many forms of art. Sure you might still have people painting or writing (mostly bad) amateur novels or whatever, but it would not be like the shift from typeset books to print on demand books, or even like the loss of vaudeville to film. It would be an enormous destruction, and a bad one.

Would it change humanity? Depends what you mean. It would definitely change society. Other technologies have before.

If you're talking about AI specifically (and you were talking more broadly about people wanting credit/compensation for their work not AI specifically; and Palworld has been accused of non-AI asset stealing) it's a somewhat different story. AI is pretty bad at many kinds of art, so using AI generation to replace some artistic work would not be a huge change. If AI got good at the types of art it is bad at...well I'm not sure I think it actually can get good at that, at least in its current form. A lot of what makes art meaningful is the way it reflects human psychology and society and current AI is bad at that stuff. It has no concept of meaning. So it can make a funny picture of what Joe Biden would look like as a hot dog but it is nowhere near being able to write an episode of a TV show. At least not a good one.

So no, AI making things like fake Pokemon (which it is or at least could be good at) would not uproot society, but to the extent it's just copying from other people's work it would serve to basically take money from people making art and transfer that money to people using computer code to copy that art.

Why as a society should we want to reallocate money that way? Why should we want to channel money from artists to copiers?

Palworld is tearing us apart.

@mach_go_go_go:

This is why AI can never replace humans when it comes to art. It doesn't have the ability to generate this depth of resonance with the emotions of other human beings. It's like watching a piece of your soul played back on screen.

@thepanzini: The problem with any of these is that it doesn't show the training data. Nor is there any way to prove that all of the training data was used with the permission of the copyright holder of those images/models/whatever.

You can say "ah, I used Midjourney for this", but that's just differing the responsibility onto someone else. Midjourney isn't open about it's training data. It could be using every copyrighted image in history and there is no way to prove it isn't using copyrighted material.

In fact, we have proof of the opposite.

So now that we know this, any material made with Midjourney is tainted. Until they purge that material from its training data, anything made with Midjourney was made while that was in the training data unknowingly was using copyrighted material without the permission of the artist or the copyright holder of that art.

And that's just one (1) copyrighted material holder. Like, if you get a generative A.I. to make any copyrighted character, it requires you to have training data from that copyright. You can't make a picture of Goku unless the A.I. knows what a Goku is and is trained to know what Goku looks like, which requires an image of Goku. And a company owns the image of what Goku looks like. That's how this works.

I know I can post a picture of Goku that I've drawn in this post, but I'm not selling this post, so Toei Animation graciously doesn't smite me from space.

I'm not saying that point just on an ethical standpoint. As I point out when I wrote the concept page, this is all super legally shakey.

Generative A.I. is going to have a place in games because company heads want there to be, but it's going to be a bumpy ride for anyone who adopts this early and thinks its completely on the level.

@devoureroftime: Haven showed a few days ago how they were using AI to turn a 2D picture into a 3D mesh model, the player will never see this. How would we know AI was even used?

Weather the AI was trained on copyrighted content will be an impossible question to answer or even find out, given every gaming studio is working on their own tools and its being baked into Unreal & Unity.

Most of the AI applications in gaming won't likely be seen.

It's definitely possible. Training data isn't some infinite, unknowable thing that's impossible to keep track of or check. There are already AI tools that exist where the people/companies making them specifically used a set of training data that they knew they had rights to. One example is Getty Images' image generation tool that they trained on their massive image library. Since it's Getty's tool and dataset they can guarantee there will be no potential legal blowback for using the tool compared to the similar tools that used potentially sketchy sets of training images. I imagine this type of approach to AI tools will be popular since it's much safer for businesses to use from a legal perspective.

@ben_h: How often can you spot assests in game bought off the Unreal marketplace compared to ones created by the developer? It's quite commen for developers to buy assests of the Unreal marketplace and heavily modifty them.

If the AI is being trained to create Pokemon or something easily recognisable but a tree, rock or background textures, not counting AI built aiding the development process. How would you know the AI building mesh models wasn't trained using assests bought from the Unreal & Unity marketplace without permission?

ChatGPT & Midjourney are two AI tools yet the latter is far easier to spot infringing copyright laws. Midjourney released in 2022 beta the lawsuit was filed nearly two years later, that's from something easily recognisable.

Okay? Assets sold on those market places are sold with the express purpose of being modified and used in games. What you described is totally fine and how those marketplaces work.

If the AI is being trained to create Pokemon or something easily recognisable but a tree, rock or background textures, not counting AI built aiding the development process. How would you know the AI building mesh models wasn't trained using assests bought from the Unreal & Unity marketplace without permission?

This sounds like some nonsense from Twitter and generally isn't how this stuff works. A person generally wouldn't be able to tell but that's beside the point (In general, usually testing whether something is AI generated or not is left up to computers since they are far, far more capable of doing so). For commercially available AI tool makers, it is in their best interest to be as transparent as possible regarding where they got their training data from because their customers don't want to have to be concerned about potential legal issues down the line. Adobe isn't going to risk their business by using legally sketchy data to train their generative fill tools.

For small teams or individuals, creating something as complicated as a custom machine learning tool using legally sketchy data they paid for to generate models would be so prohibitively expensive and time-consuming compared to just using a ready-made tool already available that it would make no sense.

@ben_h: When has the game industry ever been transparent, did Midjourney share its training data?

Google & MS have never been caught and fined doing something they shouldn't.

57:03 This is small team within Haven ~150 creating an AI tool for turning 2D images into 3D scenes, were gonna see lots of these AI tools not always commercially available or public facing.