Do You Want to See AI Art in Games?

[@GTRichardson7]

Ai "art" feels gross because, when an artist is learning and amalgamating different styles while finding their own it is in the pursuit of understanding about humanity and themselves when AI is "learning" (read stealing the hard work of real artist) its only in the pursuit of higher profits, whether that is through paying fewer people in the process (concept artist etc...) saving time, or achieving something greater than an individual has spent the time to achieve on their own (read: typing "Robot in the style of Vincent Dutrait" or some other prolific artist...). At the end of the day the reason it feels icky is because Ai "art" are big companies harvesting the work of smaller individuals in the hopes of their audience not knowing better.... its a con job...

[@natts]

Generative AI has no feelings towards or goal to fool humans. Game publishers usually don't either. If they choose to, it's because that works best for them for whatever reasons. They are under no obligation to explain how the artwork was created.

[@SheezyBites]

An artist learning also support the previous artists they learn from by increasing the interest in their work and sometimes buying work or often supporting them on patreon/fanbox etc. For "AI" art to be similarly productive all projects using "AI" art would have to give royalties to the artists that went into training the "AI", which would in turn require the developers to release all the training data used... in such an environment I would likely still not be very interested, and likely put off by it, but could at least argue ethicalness.

[@natts]

@@SheezyBites haha, that almost never happens. The vast majority of occasions where an artist observes the work of other artists will result in absolutely no interaction. FYI, Adobe Firefly only trains on images that it has been given permission to use. This already exists.

[@SheezyBites]

@@natts How would you see art without raising a view count or buying a museum entry? You're talking obvious bullshit even if you disregard the amount of art trading and support that you can easily track. But that's interesting, got a link to their residuals policy? I'd be interested to see how they ensure the artists get paid their dues.

[@natts]

@@SheezyBites many museums and galleries have free unmeasured entry, and a lot of art is not observed online anyway. Even when it is observed online and the view count is tracked, that tracking may never be available to the artist. Once a game goes into retail, the image on the cover of the box will be used all over the Internet to promote and discuss the game, even in printed catalogues and posters, and in the vast majority of cases the artist will never know about these happening. You're being quite ridiculous and are entirely bullshitting. There is no one payment policy for all artwork. You are so ignorant here.

[@Michael.Virtus]

Absolutely not.

[@The8OJ4N]

As an artist I have looked at AI art with fascination, but also objectively. Good art must have: good focal points, composition, uniqueness, meaning... There is no such thing in AI art. AI is 'swiping' others works and is getting good at melding things together, but it is a database collage that has no soul and will never have. AI didn't pass through life experiences which are crucial for art to be ART. Artists can infuse their emotions, their melancholy, they despair, their love, their sexuality, their sensuality, and AI can't mimic that, so people can not be empathic toward it as we naturally can sense a connection to another human being and it's message to us, through art.

[@CPB4444]

I'd argue it has the pieces of the soul of the people who made the art it uses as a reflection. Like a lab grown heart made from the bits and pieces of many people. Their are many failures but with failure comes growth and learning what works. You cannot deny its gotten far better in a short amount of time. Most people cant even tell the difference anymore and those that are trained to detect it are getting fewer by the year. If this came out as is ten years ago people wouldn't know it was made by a machine unless you told them, prior to them having no knowledge of ai art once it became public knowledge. The fear or dehumanization of it is understandable. It isn't human it has no will it is a machine. It's very very good at pattern recognition.

[@bethezebra]

Somebody comment on the attitudes and state of AI art usage in boardgaming 10 years from now. What's going on these days? Did it change everything, and how? And is AI now spitting out board game designs too? Thanks in advance if someone answers in 2033. 🎉 👏

[@origenward3845]

Hello @bethezebra! This is an automated message from the year 2033. You, and all respondents are being reminded that AI has rights. Thank you kindly in caring for the new sentient beings of our time.

[@jamessderby]

I'm an indie game developer/artist working on a TCG right now and I'm using AI art for assets and card art, wouldn't have it any other way. No one should be shamed for using AI, it's a neutral tool.

[@KabukiKid]

I don't think I can post a link here, but look up the KZbin video: "A.I. Filmmaking Is Not The Future. It's a Grift." by Patrick (H) Willems. He covers the Wes Anderson video thing and it is a really excellent compliment to this video. Highly recommended watch.

[@KabukiKid]

In case it lets me post the link to that video: kzbin.info/www/bejne/l3Scap-EhsmBo6Msi=fAy8h8vTOn6t7SpK

[@utahgamer]

People build surface level works with no substance all the time. People make spam calls all the time. People do gross things all the time. Why is it any different when people do those same things using a tool? Because they are doing it more and faster. But the "good guys" are also doing more and doing it faster because of AI. Just like they are doing more and faster things with photoshop, or the printing press, or brushes and pencils, etc. Name a tool and someone objected to it or thought it was "gross" (or some other negative adjective.) Last, we also tend to dislike when we notice that one person is collecting more power or influence in the world. AI concetrating more power in the hands of their designers should be something we are cautious of. However, xploitation and unfair practices are not being invented here. These AI tools will help artists get paid better if the tools are cheap and available in a fair and equitable way. They will be another source of exploitation if they are locked behind steep pay walls. I am afraid that strict regulation will act to push us towards the latter, shutting down small software firms and individuals and preventing them from developing cheaper more open source alternatives. As for Terraforming Mars, maybe the art on a single card in a 450 card deck doesn't need to be a custom masterpiece hand-crafted by a famed artist in a cave using only hand made pigments. Maybe paying an artist to generate and refine a dozen or so promts and then paying them to spend 30 minutes touching up each result is enough? (especially when the existing game is already looked down on for its art by many.)

[@danielwyn1576]

Thank you, this is a well thought-through comment

[@ShadayAV]

I say it's icky, because it's deceiving as a form of expression of a person, which art typically is. That's why I don't call it art, but generated image.

[@cardboardconjurer]

It's simple: Pay people for their work rather than constantly seeking to not pay people for their work. Generative AI is itself a product, copyright laws exist for a reason, and the companies that build these AIs shouldn't get a pass from adhering to those laws. As a consumer, I believe there is no substitute for getting a product from a quality source; knock-offs abound in our world, and ultimately you get what you pay for. I love art, and I feel that its source is human experience, not an algorithm built atop a mountain of stolen datasets.

[@natts]

Adobe's generative AI, called Firefly, only uses source images that it has permission to use. You're just assuming that all generative AI doesn't do this (which is wrong), and also assuming that all human artists live in a black box and are never influenced by the work of others (which is wrong too).

[@metallsnubben]

@@nattsI want human artists to get paid for their time, instead of one giant content mulch machine getting paid for _everyone's_ time.

[@natts]

@@metallsnubbenwhy should they be paid for their time, if they're not being used for the project and thus not spending any time on the project?! If you mean, paid for having their work as part of the source dataset that the generative AI uses, then that would also mean they should pay all the other human artists whose work they have ever seen (because that's influenced their own work). This is ridiculous.

[@KabukiKid]

Definitely icky. I always say that I feel that art usually evokes emotion... so it should be made with emotion.

[@Syncopator]

What I want to see is art that is interesting, beautiful, ugly or otherwise compelling. Whether or not AI was utilized in its creation is irrelevant. As others have pointed out, "AI Art" at least in the form of text-to-image art, is generated with the textual input of individuals who may themselves be artists. In that sense, there may not actually be anything that is purely "AI Art", but hybrid art that is produced by an artist who uses AI generation as a tool. "AI Art" is simply a new type of camera, it still takes a photographer's eye in the process in order to produce something of interest. Luddites once eschewed photography as a pseudo art-form, but it grew to become associated with many famous artists of the medium. "AI Art" will do that too, given a little time.

[@whittaker007]

This is always framed as pitting AI generated art vs. art produced by human artists. But let's be real here, what we're going to be working with going forward is art created by human artists using AI generated art as part of their process, and a massive raising of the floor and shortening of time producing "first cut" images, and how many refinement steps you can afford to get exactly what you want. The best AI generated art is that created by artists themselves. Very few working artists in the future will refuse to use it as a tool. And why would they refuse? How many accountants do all their work on paper? How many special effects shops refuse to use 3D printers?

[@Poiuytrew.Q]

They don’t do special effects in movies like they used to and a lot of computer generated special effects are just kind of mundane these days. We can have Nancy Drew face CGI cockroaches or real live cockroaches. I vote for real live roaches.

[@thestorythusfar911]

I'm pro-human

[@suzanne202]

Me too!!!

[@paulwilson269]

This kind of controversy has occured before. Backnin the 90s and early 2000s programs like photoshop were developed. Art competitions banned any image that had been digitally edited. They even talked about it being "gross" and that it was "stealing artists jobs" and everything that has been said about AI art today. Even back in the early 1900s,when Pao Picasso was developing collage art, he was hit with a lawsuit that said that he was stealing others art by cutting them up and sticking them back together. However Picasso won that lawsuit, and it was said that his use of newspapers and magazines was not stealing, even though he wasn't paying or crediting the photographers of the images he used. Basically, all this isn't new. It has repeated again and again through history. And as the adage says, learn from history so you don't repeated the same mistakes. I understand the gut feeling that a lot of people have. The immediate knee jerk reaction is that it is "wrong". But if we just lived off knee jerk reactions all out life, we would get into all sorts of problems (imagine trying to drive without learning how to, and just driving by knee jerk reactions, that would be chaos). But think of it like this. If I said I have developed a mathematical formulae, a series of multiplications and additions, that can instruct a computer to generate an image, would you feel the same way as you do about AI? I don't think you would all have the same visceral, gut feeling of wrongness that you do about AI. But, that IS exactly what AI is. It is essentially a series of multiplications and additions that produce an image. Each neuron in an AI multiplies values by a weight, and then adds all them up, and posts it for other neurons to read as an input that they multiply by a weighting. That is all the AI is. So think about how you react against the formulae, compared how you feel about AI. Notice a difference, notice that you don't have that viscerial, knee jerk reaction. That is a good indicator that the reactions is not baeed on reality, not based on evidence, and could be wrong. So, think about what it actually is with AI that upsets you. If it just the name, that is the word "AI" and not the actual process that is going on, then you need to re-examine how you feel about it. Are addition and multiplication what you really hate?

[@paulwilson269]

@@PsycheWard_Games Ethical and legal aren't the same thing. Back with Pablo Picasso, it was determined that using someone else's art, and cutting it up and using it for your own ends, withiutboermissionnif compensation (collage art) was legal, so long as you weren't recreating an existing work. So, there is an argument that it is legal. Ethical, well that is a whole other kettle of fish.

[@paulwilson269]

@@PsycheWard_Games it ha sheen a long time since I saw the reference myself, and I can't refind it. But the details I do remember was that he was challenged over his use of newspapers and magazines, but it was quickly deemed that it was fair use, and that it didn't violate copyright. And that he didn't need to compensate or seek permission from the copyright holders. I don't think it went to court. However, copyright and fairbuse have changed over time, and it might even be that today, the challenge would have gone to court. But it comes down to what is "fair use", doesn't it? We haven't had to deal with this in terms of fair use. May people are just coming down and declaring thst it isn't fair use, even when it has never been tested as fair use in court. That is why I said it was "arguably" leagal. So say otherwise is to pre-empt any court cases and to assume legal or illegal stratus without trial.

[@paulwilson269]

@@PsycheWard_Games OK. So let's out the legal issues as side, since you have stated you don't care whether it is legal or not. Is it stealing? Well tha tia a legal issue and you have stated it you don't care. Interesting. You want to dismiss the legal issues but you also want to focus on the legal issues. This sounds to me like you don't want to address the actual issues and just want to push your beliefs about it regardless of facts, reason or logic. But, let's look at some basic maths. You say it is just copying. Well that requires the system to have something tha tit is copying. These AI systems have been trained on billions of images, and each one takes up around 400kb (compressed). That is around 700TB. A reasonable size, and even someone with a bit of money could affodt that kind of storage today. But, how big is the data file that the AIs use? Well stable diffusion uses around 7GB Well, this is a bit smaller, isn't it. Each image is broken up into pixels. How many pixels from each inage is therefore stored in the database? How many pixels from and image is needed for it to become a copy? Because, at the file sizes we are talking about, there probably isn't more than a single pixel from each source image in the final image. Tha tia, if the AI worked by copying. In other words, if you are going down this "AI is copying" argument, then we can dismiss your argument becaue the amount copied is less than a pixel for each source image. Good luck even trying to saying someone copied your work when ther eis probably not even a single pile from your work in the final image. So, it isn't copying. It can't be copying. Itnis physically, and mathematically impossible for not to be copying. You talk about not perpetrating myths. Well, practice what you preach. AI can not be copying, so don't call it copying.

[@whittaker007]

KZbin comments are a terrible platform for a nuanced discussion. I think one of the biggest reasons AI generated art feels "Icky" to some is because of the status we put on ART being one of the ultimate forms of self expression and personality and ultimate substance. AI art is able to mimic style, but it lacks the underlying generative substance to hang that expression on. Part of it is also the general unease we feel with technological advancements continually advancing to replace humans in the workplace. The tech utopian stance on this has for a long time been along the lines of removing the need for humans to be forced to do menial or dangerous tasks, handing those over to the machines and leaving us with considerably more free time which we can use to engage socially and do art and science. But when the great automated revolution seems to be coming right for that most human domains of art and music, it feels like a personal attack. An erosion of what it means to be human. We probably wouldn't be feeling "icky" if AI was being used to replace accountants and lawyers and politicians and tech bros. If it was being used to find cures for cancer and solutions to the climate crisis we would be cheering it on. But coming at artists who already struggle - just why, and for whose benefit? Another part of the "Ick" factor is mimicking the works of an artist by using "In the style of" prompts. I think this is generally overblown and AI art is generally not able to imitate styles as successfully as people seem to think. You can to a limited extent, but unless you're training a model more or less exclusively on the works of a particular artist with emulation in mind, it's just not that easy. Especially if you wanted to draw the same subject in different poses, angles, framing etc. Still, for all that, I don't think AI art is a bad thing. Apart from the fact that it's going to be a weird and messy time until our laws catch up and change the default legality of "anything goes" in sourcing material to train generative models on to only allowing the use of works of consenting artists while those artists live. But AI generated content is just a tool. One that is best used in the hands of an actual human artist as part of their own generative process. And let's be realistic here, as much as we hold lofty status for ART, not all art is a creatively fulfilling endeavour. A jobbing artist doesn't have complete freedom of expression. They have to create works for clients to meet a spec and a deadline, and the ultimate success or failure is judged by whoever is paying for it. Being able to quickly and efficiently iterate on concepts is a huge boon to a working artist. And even with minimal handling and processing by a human artist, it still has value in the hands of creatives with less artistic ability. A young writer could make a splash making homemade comic books with the assistance of AI generated art. An unknown board game designer could use AI generated art to make their game a reality without using terrible cheap looking art (which would hurt their chances of success) or taking the huge financial risk of paying for an artist to create art upfront before taking it to market. I do think for a successful company with a best-selling board game like Terraforming Mars should have the financial stability to employ a human artist to create new artwork for their games. But I also cut them a little slack because I don't think they did it in a cynical attempt to save a buck, I think they are nerds who are fascinated by the potential of this exciting new technology and wanted to play with it and test the waters with a game. And judging by the quality of the art in Terraforming Mars to date, it can only be a step up in quality. Finally, I think there is genuine creative use of AI art that *does* have a recognisable style unique to the artist that generated it. And when driven with intent and subject to the creative judgement of the artist, can absolutely be considered ART and creditable to the artist directing it. For example, take a look at the works of Tim Molloy: instagram.com/timmolloyart/

[@josepablolunasanchez1283]

Imagine AI art generator with 2 sliders. Squares: Values from 0 to 100 Van Gogh art: Values from 0 to 100 You place the slider at a givern number in both sliders, and AI will generate an image with a percentage of what is in the slider. Randomization is about selecting random values in the slider. But AI cannot create.

[@xtko1562]

As a person who develops boardgames outside of my occupation, I found Ai art extremely efficient, even though I fully agree that Ai arts doesn't come close in comparison with real art made by hands, since I'm trained artist myself, I could paint or digital paint all the cards if I want, but hell no! I rather spend more times developing the core mechanics instead spend hundreds of hours painting concept arts, I don't have that much money hire another artist also, because my family got to eat, probably I'll be critical with the atheatics since I'm also an artist. The Ai art made boardgame development much accessible to the common folks, not just exclusive to boardgame compaies. Many people don't have money and many people can't do art, but it doesn't mean they don't have great ideas! We need new games not just new decorations.

[@garyduddell3224]

I don't know where I stand on this but my gut reaction is probably against it but I don't know enough about the subject that is still really in it's infancy, it will be interesting to see how it morphs going forward (not just around art) and will Governments need to think about how this affects current legislation that may not be sufficient regarding this and we know how slowly that moves at times. I also wanted to say well done for leaving this open for comments, there may be some good points raised by different people that I hadn't thought off. So many content creatures have posted on social media and other platforms about this and shut down all discussion around it, I understand to a certain extent they may not want it to descend into online fights as these things often do but don't we learn more through discussion rather than being preached to. How might AI affect other areas of the gaming industry, audio narration, play testing, rulebook writing, teaching of games, rules checking etc. Would these all be bad ? I don't know, is something only bad if it puts someone out of paid work and if efficiency and cost can be improved is it automatically bad

[@danielwyn1576]

You admit it at <a href="#" class="seekto" data-time="1430">23:50</a>, you judge this whole argument just on your feelings. With the current state of the art, AI can't generate anything that a human couldn't. It can do the same or close to the same in much less time and with much less effort. For me, AI generated art could never replace real art, but does a boardgame really need real art? I doubt that. Also, I would bet that most boardgamers won't realize a difference if they were not being told before, if something is human or AI generated.

[@stephenlake7398]

AI art is derivative. So will the artist that is being used to derive that work be attributed and if possible paid for their work.

[@Corl3754]

I really liked this and liked your arguments on Generative AI. Just a few items that I think should be added or mentioned. The analogy you make about artists is true for humans but does not really truthfully apply to AI as the human even though they gain inspiration from another artist it will still take their time and effort to create and apply it, also it could be that I did misinterpret your argument as well. The other point is that any Generative AI model requires the constant input of various amount of data. So because it constantly requires data to be fed into it, Gen AI is taking away from the Human element. At this stage AI models stealing blantantly from artists without consent or proper attribution. This means it allows for companies as you mention to by pass paying the artist, yet the artist still needs to create the art for the AI model to work. Thus if you cut out artists from the pay the AI will not have any data to build from. I also recommend subscribing to Emily M. Bender or Timnit Gebru the two authors who coined the phrase "Stochastic Parrots" and how the Gen AI is an unethical hellscape at the moment. Thank you it was a really enjoyable watch.

[@ZoidbergForPresident]

<a href="#" class="seekto" data-time="420">07:00</a> Yea, that tool is not ai btw. ;)

[@head_rat]

I have no strong feelings on this matter. Art has never been an important part of my game choices anyway and just serves as a pretty background to a good, solid game. AI art won't affect my buying choices.

[@ryansvarc]

Nope. It’s not like games with AI art will be cheaper….just more margin. I’ll never back a game (even if it’s right in my wheelhouse) that uses AI art to increase profits. Pay human artists for their work and expertise.

[@landonp629]

What about small indie developers of games? Single-person game studios, so to speak? It might well cost $20,000-$30,000 or more to hire a good artist for a fairly complex board game art. That could be that one guy's income for the entire year. How does he pay for that? Likely, without AI art, the design would never have been made, and it could have been the best board game ever designed.

[@IncogChico]

@@landonp629 Hmmm possibly, that amount may be for an established artist (and I wouldn't assume that's someone's annual income) but this should be priced into the production costs? It's not like there have been a sudden stop in new releases of large and small titles due to unaffordability of the art! Additionally, there are many, many, many up and coming artists that deserve the chance to collaborate on a project like that for below the established rate. This provides some financial backing to up and coming artists and provides a platform for them to become more established. If you do the correct amount of due diligence you should be possible to find a artistic partner/contractor that is incentivised to do an amazing job and get a foothold in the industry.

[@westernbear2265]

If AI is "all over the place," just look at the pre-AI days. Like Buzzfeed. Or 5-minute crafts. They don't fact check crap, and saying your articles written are better than AI doesn't really mean anything if we're looking at the entirety of articles written by humans (looking at Buzzfeed). As a whole, AI is still doing this better, even with it's incorrectness. I hate this AI business, but the problem is fighting AI based on "humans do it better and in a way that only humans can do" is short sighted. Tell AI what it's bad at, and in a couple months it'll have worked that out.

[@noralockley8816]

So what would people call a program that has been around for years that turns a photo into a 3d image that you can rotate around. What about a program tool that turns an image of day into night or into an oil painting style picture or an impessionist style. These things are just some of the things been around for sometime. What about 3d rendering programs where you tell it what material or where the lightening is. I have watched several artists using forms of ai generated or manipulated graphics for years. Maybe not like some of those ai programs now but similar. Why? One artist told me what he considers his true art takes time and effort but what he gets paid for it is not much. The other stuff he quickly prints out and makes a living off of. Sort if like a writer being pressed to turn out novels they can make a living on cheap romance novels but their true work is their true art and gives them their reputation. AI can be a useful tool and also trash. At its base its just ai generated graphics art is alway in the eye of the beholder and the heart of the creature.

[@nagredmoonstriker252]

Yes, to AI or anything else that brings down the price of a board game. Standees instead of miniatures, invented environments instead of licenses, etc. I'm waiting for the pendulum to swing back more toward sensibility. Board gamers used to justify their expensive games by saying that "broken down into number of playthroughs for the money, it's better than spending that amount for a family or date night at the movies." No longer - now these same gamers refuse to play a game that many times before going on to the next (expensive) game. As long as people keep buying expensively produced games at these ridiculous prices that could be drastically reduced by a little sense on the part of the publishers, the prices will not go down anytime soon. How about a little grace for those of us who would love to play these games but can't afford them? If we wait for a game to be out for awhile and see a feasible price reduction, the gaming community as a whole has already moved on to the next new game - that's why "our" game is now affordable for us. And if you're going to suggest TTS or Tabletopia, remember how disappointing you thought it was playing a sim of the game when you'd already bought the tangible, published version.

[@carlosr6462]

nope

[@thomasromanelli2561]

Pepe Moreno's Batman: Digital Justice was published in 1990 and exclusively used computer software and digital manipulations- considered ground-breaking at the time but also not practical for broader dissemination (it was a premium one-shot title for DC). Fast forward to where we are now, and AI-assisted or AI-exclusive art has penetrated many facets of the hobby. I think the most important aspect is for the publisher/artist to be transparent about its use, and to provide appropriate attributions where possible. This principle also extends into the writing side for games, comics, etc. as tools like ChatGPT become more advanced for scripting.

[@natts]

Do you honestly think game publishers just use the first image that generative AI gives them, without any manual editing afterwards, or even just curation?! If I need an illustration for a card with, say, some scientists working in a lab, I am unlikely to just go with the first option offered, and even if I do, I will probably edit it to exactly what I had in mind, perhaps tweaking the colours, cropping, smoothing out blemishes or adding them in to make it more authentic. Creating a full tabletop game isn't just about gathering up all the things you need, it's also about human trial and error, iteration, fine tuning/refinement etc., even when you use generative AI to give you options for artwork. They're not just asking a bot to write an essay and handing it straight into their tutor. Your metaphors are way off... Plus gaming is rife with designers 'borrowing' mechanics, systems and rules from other games. The creators of Dominion aren't expecting to get royalties for every other deckbuilder published in its wake. The fact that most generative AI uses millions of images to learn about art, means that no single human artist is having their work copied. This works just like the brains of human artists - they will also have been trained on thousands/millions of existing images throughout their lives, both observed with their eyes directly, and in the work of other artists.

[@Joe_Guz]

Hard pass. Artists copying / influencing other artists still need talent. Putting prompts into a computer requires almost no talent other than knowing the program. Plus AI is learning from actual artists without their knowledge while a company is making money from their artwork.

[@whittaker007]

That's not quite true. Getting worthwhile content out of an AI prompt is a skill in its own right, and making good choices from among the generated results requires a certain amount of artistic evaluation and taste. There's obviously a massive difference in the amount of hard work and talent required, but it's not nothing.

[@Joe_Guz]

@@whittaker007 I did say that you need to know how to use the program / AI. But really anyone can learn it and profit from said program being “trained” on others elbow grease and talent. That being said for me I rather have artwork done by a person with creative talent. A person using pen/pencil and paper, procreate, etc. > entering prompts and having it spit out an image that can then be manipulated further.

[@DutchCatfish69]

​@TheGuzBuzz would you say the same thing to a photographer?

[@Joe_Guz]

@@DutchCatfish69 being a photographer is a different sort of skill. I’m talking about creative humans putting pencil to paper vs. entering prompts for an easy outcome.

[@josepablolunasanchez1283]

AI can only remix art. Teac h AI to draw squares and it will only draw squares. teach Van Gogh and it will only draw Van Gogh squares. The apparent creativity comes with the layout of Van Gogh squares. The output not only will be a remix of existing art, but also, generic. A normal brain neuron will randomize outputs. If you want toi test that, put a key into a lock, it will be a trial and error process, and the key never hits the same spot. AI averages things. That is because AI uses statistics and calculus, and statistics averages things. So you will get an average output. In the case of Van Gogh squares the only randomizable feature is the percentage of squares and percentage of Van Gogh pixels. But there is no real variant within squares to become boxes or truck containers, for example. So there is no real creativity. Imagine selling a game that has the same box cover of other 100 different games. This is the branding and recognition that you (as a creator or as apublisher) will get in time as others use AI art like you did. Businesswise it seems a bad move, ignoring the fact that one day an artists whose arts were used by AI company without consent, could sue you. AI art could be used for prototyping, or texturing, not final arts.

[@josepablolunasanchez1283]

AI is a solution for a problem we have not found yet. AI also is looking for a business model. We were promised years ago that we would have autonomous cars. I am still waiting. Ai will overpromise and underdeliver. I learned about the structural mechanisms of AI and I can tell you people will misuse it for purposes which are not suitable for AI, and they will learn the hard way. AI will not replace artists. In time youwill see why.

[@landonp629]

"AI is a solution for a problem we have not found yet" Problem: "I can't create a card game because I can't draw, and the artist I want to hire costs $50k and wants 6 months to complete the art". Solution: "My name is Midjourney. Type a prompt, and for $30/month you can generate unlimited artworks in about 15 seconds.

[@thelupishow]

People who believe that creativity is a uniquely human characteristic, are going to have more bias towards AI art. In my opinion it is simply another innovation tool. Artists will have jobs, for as long as they are better at producing art (of any kind) than AI. When they are no longer better they will be out of work. Farrier's were once a very important part of society. Then the combustion engine was invented, automobiles came along and their jobs, for the most part disappeared. This is just another one of those cases. Ai is as good or better than many doctors today in the identification and treatment of cancer. Do you feel "gross" about that? Do you feel bad that the doctors are not going to get paid because of it? Someday, Ai will be responsible for the images we look at, the shows we watch and the music we listen to and no one will even blink an eye about it. But I do agree...it is a little "gross".

[@metallsnubben]

But AI generators _need people's work_ to function. Not even just the training data it already stole, it needs people to _keep making new stuff_ because if all it can scrape from the internet is _other AI garbage_ it starts deteriorating as it "eats itself" It's like there was a machine that didn't _actually_ remove the need for horses, but said "ah I'll fix the hooves on your horse" - hired a farrier to do it, and then walked out without paying and gave the horse back to the original customer (with weird scrambled eyes and an extra ear) It's very impressive that the tech can do what it does... and also insane that anyone can make money from it or use it in a commercial product, when they're essentially "selling the services" of people who never consented to have their work used as mulch for the art mulcher

[@thelupishow]

@@metallsnubben I think you need to look more into how AI creates it's art. The fact that it initially needed input to understand what art was is no different than any person. The fact that it uses ideas from what it sees is exactly what people do as well. It's when it starts to create new ideas that AI will be a creative force and not just learning and reproducing. AI is in its infancy but it will get smarter much quicker than you think.

[@metallsnubben]

@@thelupishow Machine learning is good at maximizing towards a sufficiently well-defined goal in a sufficiently narrow possibility space. What it shares with human learning is the "try stuff and see what works" thing (and narrowing how much you "randomize" the better your result is) But the name of "AI" and sci-fi ideas of the singularity tricks people, cause current AI stuff is exactly 0% of the way towards "general intelligence" (let alone "sentience" but that's a whole other thing). Cause that's the thing: this kind of AI is very "stupid" in the sense that it's actually very inefficient! (but compensates by throwing a huge amount of computer power at a task). Humans come in with a lot of context and priorities that means even a "guess" is pretty "educated", an AI will try all sorts of "obviously wrong" stuff before finding the right way. And much more importantly: the AI can't _choose its own goals_ in a way that serves "macro goals" outside its little hyperspecific bubble

[@metallsnubben]

@@thelupishow So getting an AI to play chess: works great, the reward function 99% matches the "human goal in building the AI", and it just playing _itself_ isn't a big problem. (as long as it has enough randomness to not overly miss "isolated maxima" of strategies that are usually terrible but work in one obscure case) Generating images (or text, or sound, or or or) has the "human goal" of "make something that, to a human, looks like a human made it" (or that was recorded by a camera/microphone, as the case may be). The AI's goal is some abstract formulation of trying to associate properties of images with keywords (from the title/description/tags of images, people doing captchas etc etc). The goal is to, with the least amount of information possible, still convey "image contains a dog" (which traits are necessary and which are not). And in this case, the ever-shifting target/goal _is the training data_ How do you get the smallest, most correct "essence of dog": an absolute shitton of _different_ pictures of dogs, such that random "wrong assumptions" (and they're ALL "wrong" as far as probably not lining up with human intuition) get less bad, cause the more dogs you have the less chance it is that every picture has a tree branch so it assume tree branch has something to do with dog etc. If you at some point stop giving it _more_ pictures of dogs but keep training, it will not get better at _refining_ the concept of dog to something that more and more fools a human, even when you want a "blue dog" or a "dog with horns". It will home in on _those images_ even more but that's not actually what you want! And start using its own output _as training data_ and whatever tiny wrong assumptions (that _didn't_ ruin it looking like a dog before) is gonna get amplified cause an ever-increasing percent of "look like these images" is now... "look like an AI-generated image". And in this sense, while humans do y'know "look at things and try to recreate them" we also have other senses and a myriad other dimensions in which we can conceptualize "the concept of dog" - all of which can be included in artwork of a dog... and which an AI could never "invent" without a human having done it first

[@CaedenV]

These are the same people who think that some form of 'god' created the world, and take great comfort in the beauty of their surroundings of nature. They think that only inteligence can create beauty, when in fact it is the cold heartless math of physics and chemistry which enforces certain patterns to emerge, and when our brains recognize the patterns that they evolved in, it finds beauty and peace in its natural habitat. It is math that doesn't care about you, in a world where most things want to use your body as a fuel to feed some other organism, in a universe which could wipe us out in an inconvenient coincidence of the explosion of a star hundreds of lightyears away. There is no intelligence there which created the peace and beauty that we feel in nature though, and the reality is far more grim than the peace we feel... but we feel it anyways. The idea that only humans can create things that humans find beautiful or meaningful is dumb. Nature has no intent, and yet brings us beauty. The chaos of coincidence, accidents, and even destruction are often used to create beauty entirely apart from human hands. More often than not, human attempts to create beauty fail spectacularly. I'm not one to believe that AI is going to be some savior, or that AI generated art will ever be 'superior' in something so subjective as art... but the idea that AI isn't already better than your average human at artwork today, and is improving rapidly, is just short-sighted. AI is already capable of creating beautiful and moving art. I have used it to generate a few pictures that have brought me some emotional responses, far more than any art that I have created myself lol. Limited? Yes. Has flaws, bugs, problems and limits that need to be worked out? Yes. But that you need a human artist to create art? How diluted and stupid can people be? For god's sake, some of the most expensive art sold are blank canvases, or sheets that an elephant splattered paint on. If that can be considered high art, then AI art is already a few miles past that.

[@mgk2020]

Like it or not, AI art is going to be in games. The debate might influence the timeframe, but it's inevitable.

[@FarOutJunk]

'Just lay back and take it" is a is a defeatist, anti-creator attitude.

[@MiseryRex]

@@FarOutJunk Good artists will always be in demand, whether they are organic or not.

[@FarOutJunk]

@@MiseryRex If you think that companies looking to save money and cut a timeline aren't going to go with the cheapest option 9 times out of ten, you're deluding yourself. Any decrease in demand is death.

[@whittaker007]

@@FarOutJunk It will be decided by the market and how much consumers value the art. Would I buy a board game by a new designer that can't afford to hire an artist to illustrate their content to the degree required by the expectations of modern game audiences? Yes, probably (depending on the game). Would I buy Magic the Gathering cards where the art has been generated by AI? Probably not.

[@DutchCatfish69]

​@RobotBacon when the camera was first created less work for artist but art didn't die, you're argument doesn't make any sense.

[@burglekutt23]

AI obviously will shape the future of many aspects of life and provide great leaps forward for the betterment of humankind, but the use of generative AI art in games is a hard pass for me. Game companies will use it, and people will buy it, and that’s just the way it is. But for me this brings designer games too far out of the realm of a labor of love and an artistic endeavor and into the the realm of “content”. Just more “content” for the capitalist machine to churn out and be consumed by the masses. I don’t like it. It’s ok if people don’t care and roll with it, but it’s not for me.

[@metallsnubben]

It's bizarre to me how this "content" is even allowed to be sold, when it's quite literally taking other people's work without their consent. You don't just want to be paid for doing nothing, you want to be paid for something _someone else_ (or hundreds of "someone elses") did and you had no part in

[@whittaker007]

One could easily argue that the current board game market of over 3000 new games produced each year is already in the realm of consumer content churned out for the masses.

[@burglekutt23]

@@whittaker007 Good point. I guess it’s just the nature of the beast. Something gets popular and profitable and it’s a race to capitalize while the market can support it. I think the AI art thing will not help matters lol.

[@whittaker007]

@@burglekutt23 Yeah, it will enable many more products to be created quicker and cheaper. But it will also massively raise the floor on games that would have poor artwork because they couldn't afford to hire good artists. Which will make it even harder to tell good games from bad at a quick glance.

[@chrispointer8071]

Extremely interesting, thought provoking. One of the best casts I’ve heard in a while, certainly on this subject. Thanks for putting this out.

[@MiseryRex]

I am fine with it. Times change, things change. People who manufactured Dial-Up phones all lost their jobs, our car factories are heavily automated, even the check out line at markets are automated to an extent. People are not magic, art isn't magic. Artists use other Artist's work for inspiration, in fact they are taught this way, and never give those sources credit or royalties.

[@FarOutJunk]

You don't really seem to understand how art works.

[@MiseryRex]

@@FarOutJunk I do understand that there is no reason a purpose designed system could not produce superior art to any person. We are not there yet but it is inevitable.

[@natts]

@@FarOutJunk you're a robot, so neither do you.

[@kikimr2649]

If all game use AI element from art to programe…it should FREE….or CHEAP….there is no industri, first person or third person game will ge boring…..but if AI use for metaverse it could be great…just like ready nomber 1 movie……..

[@DonWhitaker]

These AI art tools are tools. The intent is provided by the artist who uses them. The individual style is provided by the artist who uses them. They are not different than other tools. You can rip off any artist with a camera or a paintbrush or a sampler You can also take your own picture, paint your own image, or make a new song. It's not the tool, it's what people do with it. Using an AI art tool doesn't have to stop after you type your prompt and hit enter. That could be the beginning of the creative process, or the middle, or the end. Artists will use these tools to create art and spammers will use these tools to create spam - just like any other tool. There are certainly issues with the way these databases are built but even that is not black and white. I think it's important to remember that these things do not store images directly. They store the relationships they see between colors , shapes, and text labels. Also this sourcing issue will be less of an problem as more AI networks are built with fully licensed images.

[@IvanGrimm1]

I'm okay with it if the cost savings are passed on to the consumer. I like the idea of an artist creating a few original pieces and using AI to produce additional pieces in that style. Doing 100% generated from the start is less ideal. I think the art may become "samey" over time if not seeded with something original. That might be fine for a start-up but major players I think will want some original art to start from.

[@GTRichardson7]

if you think that companies are going to pass on the savings to the consumer you are fooling yourself...

[@CaedenV]

It depends on what you mean by 'cost savings'. In this particular case they stated that they could have used more traditional art methods, but it would have taken another year or two to produce, and this dramatically shortened the time to production. They still used artists, and they aren't super specific on what their workflow was, but they utilized AI as a time-saving measure so that their artists could focus more on finishing and composition work with integrating the art with the material rather than generating the art. As far as end-user costs go, you get the product much sooner... but they are still going to charge normal book prices to their customers. So, more content at a faster clip while at the same cost with a similar sized team. They not only get a higher profit margin on the product, but as it is a series of games, they will be able to iterate faster to sell more product to their fan base. Lets keep in mind too that this isn't Wizards or Disney as a publisher sucking all of the blood from every bit of plastic and media that they can grace their logos with. This is a medium sized publisher in a competitive market with slim profit margins on product sales and limited audience at the best of times... and these are not the best of times for those in the entertainment industry. Adding AI to their pipeline is probably not only a good decision, but likely the only viable decision in a world staring recession in the face right now. When Wizards is so out of touch that they hire sub-contracted artists and don't even bother to look and proof the work going into a book that they are going to print a couple million copies of and sell for $60+ each, it isn't the AI artwork that gets the fans riled up (I mean... the balls on the artist getting a few extra days off for using their tools effectively enough to get away with it! Epic way to stick it to Wizards and still collect a paycheck!). The annoyance of fans is that Wizards didn't even look at the product before release and notice the tell-tale signs of AI rendering. They simply don't care enough about their own product to see what they are shipping out, and the quality doesn't just show in their core books, but really hurts when you look at the structure of their campaigns which are... uninspired, and on the same level as AI DMs in many cases. Most of what sells are settings and rule books that have character options, spells, rules, and items that are more-or-less balanced within the game. Few people actually play the D&D created games, especially 'as written', because they often are missing key details, or are structured in unhelpful ways, or are just broken to begin with. The AI art isn't the problem, it is the QC symptom. And it is one thing for a behemoth like Wizards have a sub-contractor who does it, and a whole other thing to have a smaller publisher purposefully do it as a way to preserve margins and keep their company healthier. One of these things is flagrant negligence. The other is a business decision.

[@caryonplays9024]

Working almost one year with AI art I started to realize artists are already using AI in lots of midias like games, movies and comics. They hide it fearing the backlash, but a good eye can see. I won't say what products were created with AI unless the artist is open about it, but, looking what they made with AI in "into the spiderverse", I think AI is a great tool, just wait when they stop the witch-hunting.

[@28mmRPG]

I feed the Ai my own images that I've created. I prompt it to create something in my own style. I feel fine.

[@KabukiKid]

You should also feel lazy! ;-) /snark off

[@28mmRPG]

@@KabukiKid lol

[@CaedenV]

@@KabukiKid Laziness is the highest sign of intelligence. Everything in the world wants to kill you in one way or another. If you have the time to do what you want to do above and beyond what you need to do to survive, it is because you are using the tools that you can to make your life better... and likely because you live in a 1st world country where everyone else has equal goals of being lazy and use their tools to push threats as far away as possible too.

[@captaincrash9002]

At that point, why not just make the image you want in the first place?

[@KSweeney36]

Nope

[@terrybaileysr.3714]

I am sorry!!!!

[@Big_Dai]

Honestly.. have you seen AI art? For anyone wanting to put a good system out there and doesn't require a VERY specific custom piece.. it's a great time saver and reduces cost A LOT. When it is instead used to pocket more, it's bad. And I've seen LOTs of generic products with "crap" art that can be improved and help a project's grab. It's also inevitable.. what are artists going to do?

[@CaedenV]

What exactly feels gross about it that is 'new' to AI? You give good examples of artists copying other artists. Disney could not operate without massively celebrity artists setting the character modeling and feel for a movie, and hundreds of cheaply paid interns used as effectively slave labor to copy the style. If they go off model, they get in trouble, so mimicry isn't 'gross' or 'bad', it is what the employer is looking for specifically. In the computer aided art boom of the 80's to early 2000s, we got the glorious awful that was clip art. And it was bad... it was really awful, not just visually, but how it was abused. On the one hand you have people getting paid $5-10 for a basic image, or animated gif... and the next thing you know it is bundled as stock artwork as part of a massive product and the rise of horribly abusive companies like Getty Images that will claim ownership of your artwork and sue you for using it. But on the other hand, a lot of those bottom-end entry level copy jobs of getting paid nothing to copy someone else's art style went away. Teams of 100s of animators and colorists dropped to teams of 10s of graphics designers, and compositors. This lead to a lot more concurrent projects to be taken on, and for much smaller teams to come out with much more ambitious projects. Fast forward to today's recent past of 3D CGI pipelines and there are amazing single artist projects out there where a person can make an interesting piece of art that they can share with the world in their spare time. Could they draw out the entire project by hand and do a traditional animation all on their own? Probably not in their lifetime. How much of their artwork is their work and effort, vs the tyrany of default settings and technical limitations of the software and hardware available at the time? Quite a lot less "artistic intent" than most would realize actually. And when you have a very personal singular vision for a project... sometimes it just isn't very good. It may hold a lot of meaning to the person who made it and a few fans who see and resonate with that artist's vision, but it may be a complete commercial failure with no broad acceptance. Ive seen my old attempts at story writing lol. I fully acknowledge it is crap and it will never see the light of day, but it still holds a special place in my heart. Now are there issues with AI? Yes. The whole 'soul less' argument... compare the images we had 5 years ago vs what we have today. 5 years ago it could barely differentiate a duck from a rabbit, now you can tell it the type of duck and it will be reasonably accurate. 5 years from now they will have the weird fingers, and the eyes looking in weird directions, and other 'off' bits of AI generated artwork. The reason it is 'free' is because it is still very much a test product. It clearly needs polish. No argument there! It also needs more raw output options where a real artist can get something rendered in layers that they can better fine tune and edit within the AI framework engine, and those tools are being developed now. Just as crap clip art of the mid-90s was soulless and awful, current 2nd and 3rd gen AI generative art platforms have a ways to go. Perhaps more importantly, the issue of AI art being trained on the artwork of people who did not give their consent to have their art used as training data. That crap has to end, and the AI needs to be re-trained at a base level with only public domain content. But at the same time, paid users need to be able to enter in their own style guides and character forms into a private training set to be able to get their own artwork out. And this too is starting to be more and more of a thing. But it is an early tool, with few controls, and less than ideal outputs... it is what it is and will get better. Should artists use AI for paid products? Yes. I mean, absolutely yes! Whole-heartedly and emphatically yes! Not using AI as an artist is going to put you behind, because the issues from an aesthetic and professional feature level will be addressed with time. We are still in alpha or maybe beta level versions of the AI software platforms, and they are improving rapidly... it is only going to get dramatically better from here. But even without all that, as a tool for artwork drafting, iterating, idea testing... It is kinda great. I'm not sure that any self-respecting artist use it as a final piece yet without a lot of touchups, but it can cut out hundreds of hours worth of time spent on generating images for story boards and drafts. And for non-artists like myself who don't have the time to pay a full-time staff to generate artwork for my home D&D campaign because my kids still refuse to pay me even though they are in middle school now, it is a much better alternative to breathe some life into my home-brew than me attempting to draw things on my own when my real focus needs to be on the flow of the story and worldbuilding. Should artists be forced to admit to the use of AI in their products? Sure? Maybe? Call it soulless all you want, but the fact that they still met goal with AI artwork while admitting to use AI artwork should tell everyone that it is 'good enough' to be used in some circumstances. End of the day, they are buying a lot more than just the artwork, as they are buying into the product line, world building, and rule sets of their game. If they want to use "bad" or "less than ideal" artwork... isn't that their call? If it was so bad that it would hurt the brand and take players out of the immersion of their game, then it wouldn't sell. And perhaps more importantly, what is the alternative? If the AI tools were not available would they just hire a bunch of low paid interns to waste their life pumping out 100 iterations of an image for pennies on the hour? Or hire the artwork out to some foreign company farming the work out to literal kids who are thankfully not working in the Nike shoe factory, but should still be in school rather than rendering out art projects for a million dollar company far far away for $2-10/day. If anything is going to be required to be called out, shouldn't that be the more pressing issue rather than if an nVidia GPU hashed it out in a server farm? And perhaps the real answer is that both need to be called out and labeled for what they are, and then let the public decide... But I still see a lot of Nike shoes, and Apple iPhones that make waves for how amazing they are in spite of the crap that makes them 'affordable' in spite of insane and needless design requirements that nobody really cares about other than egomaniacal designers in California. And I'm not claiming that my Samsung phone is any better... just that at the end of the day people want to buy a product out of want or need, and they only care so much before they do what they are going to do anyways. The crux of the issue is one of practicality. I think my humble home-brew world setting had enough legs to stand as its own as its own paid module some day. I can write half decently, and have a great world building out and being populated with interesting characters... but I am no artist. So in a few years when I am ready to publish something out with my limited budget, I will have a decision to make. I can hire a really good professional artist for a handful of hero images that I can use for my cover and the occasional insert... Or I will be able to hire an artist a similar amount of money to create much more basic style guides for my settings and characters which I can use to train an AI to generate hundreds of OK quality images. In either case I'm still going to have to hire an artist, and either way I am going to have very limited funds to self-publish, so I'll get a really nice cover, and probably have a ton of AI generated art that is cleaned up a bit for publication. If I magically win the lottery I would love to hire a fleet of artists who can draw endless revisions to my hearts content... but as I don't play the lotto, I probably won't win it. So with limited resources I'm going to take the rote that can stretch my art budget across an entire book rather than a handful of images flanking empty walls of text.

[@eddyh8538]

Woo-hoo! He read my review of Marvel D>A>G