My Views on AI Generated Content

Forum News
1

So now onto the one I am personally not looking forward to, but it’s got to be done. Due to how controversial this is though, I want to let everyone know that we are going to be a lot harder than we tend to be in terms of moderation and I am going to set some ground rules and some guardrails that if violated we are just going to delete your posts with no warnings.

  1. Stay respectful while debating with each other - I actually think there are quite a few people on both sides that do this very well, but I also know this is a very hot button topic and some people can not seem to control themselves so if you are feeling heated take a breath before responding.

  2. Don’t argue about the legality of LLMs and generated content - While I may personally believe that content generated by LLMs should fall under derivative work currently the courts are more leaning towards fair use. If you do not like this here is not the right place to argue it. Talk to your representatives and argue with them to change the law. To make sure we are all on the same page here is the general state of legality of LLMs for their major use cases at least in the US:

    1. Code - LLMs where mainly trained on open source repos and code snippets whose licenses allowed for such use. While it can be argued this is another abuse of open source culture by larger companies, even in the more anti-ai tech circles the legality of it was never really in question.
    2. Writing & Text - With the ruling in Bartz v. Anthropic the use of writing in training LLMs is being considered transformative enough to fall under fair use. Lawsuits are currently pivoting to focusing on if companies acquired the writing legitimately or through piracy. It should be noted that Judge William Alsup’s ruling has been criticized by some other judges and other lawsuits are trying to challenge this, but atm Alsup’s ruling is generally accepted.
    3. Art & Visual Mediums - Fair use arguments have not been decided for visual mediums and art but the ruling in Bartz v. Anthropic as well as the other lawsuit that finished up around the same time are considered heavy blows against those arguing against AI. It’s still in the air but is much more likely training on visual art and other visual mediums will also be considered fair use. Most larger lawsuits like the one with Disney and Universal suing Midjourney focus more on the replication of their copyrighted characters then arguing if the work in general is fair use or derivative.
    4. Audio & Voice Overs - I have not heard much movement here as of yet. To my knowledge at the time of writing unions have moved quickly to try to protect voice actors and other actors from having their likeness copied but that is about all I know atm.

  3. While it may be legal I think most of us can reasonably agree that it was not handled ethically - As with code, while what the AI companies did was legal it is generally considered another abuse of the open source culture by larger companies. With everyone I talked to it seems like that most everyone tends to be in agreement, whether they support AI or are against it, that how most AI companies have handled LLMs up to this point has been dubious at best and out right unethical at worst.

Also, Krod suggested it might be good to throw in a few definitions here since I throw quite a few abbreviations around.

  • AI - Algorithms meant for allowing computers to process complex problems or perform complex actions. There is a very wide array of AIs, but in this context usually refers to a class called supervised machine learning algorithms and is used as a shorthand for LLMs in general.
  • LLMs - A type of machine learning AI specializing in understanding human language. Primitive versions have been around since ~2015 usually acting as translation tools until GPT-3 was released around 2020. While LLMs generally only deal with text generation it is commonly used as a shorthand to describe any form of generative AI.
  • Generative AI - Refers to any class of AI that can generate content. While usually used to refer to recent neural net based AIs like LLMs a wide class of different algorithms including common procedural generation algorithms and game AI systems can fall under this classification.
  • AIGC - AI-Generated Content. Used mainly to refer to content generated by advanced models like LLMs and distinguish it from more common procedurally generated content.

With all that setup lets get right into it.

My Personal Opinions & Views

So before I get into talking a bit about my own personal thoughts on AI generated content (AIGC) & LLMs/Transformer models in general, it is going to really sound like I am going after the people with an anti-ai view point. The main reason for that is out of the two sides, the anti-ai side in my interactions thus far tends to have a larger radical fringe and is for sure a lot more vocal and loud then the pro-ai side of the argument tends to be.

That all being said, it’s not my intent to lump everyone with an anti-ai viewpoint into that same group as those people who are just fanatics. There are plenty of people I know with anti-ai views, my wife included, that have legit reasons and can respectfully express them without resorting to witch hunts. But I think those same people can agree that that fringe has become so vocal, so aggressive, and so fanatical that it is becoming a problem and is actively hurting their side of the argument just by association. After all, when quite a few of the people I talked to who identify as being very anti-ai feel sick by what they are seeing this radical fringe doing to try to impose their will on others it’s a problem.

So at points in this talk if I go on rants about behavior on the anti-ai side, such as spamming my DMs non stop to try to force me to ban AI or harassing people who even experimented with it because they are “undesirables”, please keep in mind I am not talking about the majority who I am sure are good reasonable people. I am talking about the very vocal minority that is giving that side an overall bad name.

For or Against?

I know some of the people I have interacted with have called me an AI Bro or that I was blinded to the reality of the tech by its potential; which always made me laugh because the higher ups at my work have the exact opposite view. At work I have been publicly berated and even on one occasion threatened with being fired for urging caution and pointing out the short falls of LLMs and especially how they wanted to apply them.

But I do know I can play my cards close to my chest, usually unintentionally. I tend to try to remove my own personal opinions during discussions and play devil’s advocate unless someone asks me directly for my own personal opinion on a matter. This is mainly because I try to dissect peoples viewpoints and reasonings to see if I can get any new information that can help inform my own, though I know for some people this can make me difficult to talk with.

So with that, let me state that where I stand personally on AIGC is neutral with a slight lean towards the anti-ai side of the argument. I do think that the tech can be a very helpful tool in the right contexts, but at the same time it is very easy to misuse, can be unreliable and hard to predict, and its weaknesses are hard to spot for those not familiar with the underlying tech.

If You Don’t Like It, Why Do You Allow It?

In general I have 3 major reasons that I choose to still allow AIGC to be posted and discussed on the forums.

First, In general I think putting our heads in the sand and trying to ignore the issue is just a bad idea in general. I see a lot of spaces out right ban even the discussion of AI which personally I think is just foolish. The fact of the matter is the genie is out of the bottle and while trying to ignore it may make us feel better in the short term, I think harms us more in the long term.Trying to ignore the issue just allows others to make decisions for us, and we shouldn’t be afraid to speak openly and honestly about how this tech can both help and harm.

Second, simply I don’t think it is my place to dictate to a dev how they should build their games or what tools they should use.When talking to someone who was trying to convince me to ban AI I stated this as one of the reasons I am hesitant to do so and their response surprised me. They said that those in a leadership role in the community like them and I have a moral duty to dictate what is right and wrong to those under us.

I deeply disagree with this on a personal level. First, I don’t feel like I am in any sort of leadership position in the community anyway, but more importantly I don’t ever agree with a moral imperative to dictate the actions of others. I always felt it is better to guide than to control, and I do my best to practice this. When I hear someone wanting to use AI I try to reach out to hear their reasoning, make sure they understand the risk, and even try to offer alternatives if I am able to do so (like teach them how to code for example), but ultimately it is their decision and if they feel the benefits outweigh the risk, who am I to question that especially if it was an informed decision.

And then finally…

Does the Use of AI Invalidate All Other Creative Work in a Game?

I have had quite a few people on the anti-AI side of the argument claim that any use of AI invalidates the work as a whole. This argument I think is fair in monomedia projects. I have a few issues with it when it comes to games.

  1. More of a personal pet peeve but for a fair number of people that make this argument, they don’t mean “any” they mainly mean visual AIGC as when I prompt them about writing, music, sfx, programing, ect I usually get reasons as to why those don’t matter or should be considered “reasonable” exceptions. This makes the argument in general feel disingenuous to me.
  2. The argument tends to ignore the fact that games are not monomedia projects, but larger multimedia projects that require a large range of skillsets to pull off.

To me while point 1 is an annoyance, point 2 is the main reason I have a hard time feeling it’s justified to write off an entire game when it uses AIGC. If more games were just fully AI generated this question is a lot easier, but the reality of it is that within our community it’s more common to use AIGC to supplement missing skillsets than to create a full game with just AIGC which makes it alot more sticky.

For example, if a writer decided to add in AI generated images into their text adventure to give it a visual component does that invalidate the work they put into their writing, programing, and design of the game? Or let’s say an artist wants to make a VN but doesn’t know how to code so uses ChatGPT to convert their writing into Ren’py scripts, does that invalidate all the work they put into their art and story?

To me, I think jumping straight to any use invalidates all creative work put into a game is just too extreme of a jump and ignores the fact that a lot more goes into making a game then just one specific skillset. And also, this question will get harder and harder to answer as AI enabled tools continue to work their way into common tools that many devs use causing devs to make use of AIGC without even realizing it.

Is This Why AIGC is not Banned from the Jam?

For the most part yes, though not the only reason. I think there are a handful of people that think a game Jam is a pure competition where everyone is trying to make a complete game in a set period of time. The thing is Jams and Hackathons are meant for practicing rapid prototyping skills, to build a Proof Of Concept, or a small vertical slice to demonstrate a concept or experiment with some off the wall ideas. Due to this most Jams expect and even encourage participants to take shortcuts since dealing with that tight time constraint is such a big part of the jam.

Due to this I tend to classify AIGC about the same as assetpacks in my head. Personally though, I do generally agree that some one shouldn’t be trying to sell a game if it’s using AIGC straight. That being said, I don’t think it’s an unfair use of AIGC for prototyping and concepting with AIGC acting as placeholders until it can be replaced… And while some people have made some good arguments about how they feel the flexibility and the fact AIGC can output a large quantity of content makes it a tool that is unfair to use, I personally think that the stylistic disadvantages of AIGC tend to outweigh or negate that advantage and our data suggests that quality is generally appreciated over quantity.

But in general my views on not wanting to dictate what devs should dom the fact no one has as of yet been able to put forward a strong argument as to why we should treat AIGC different from assetpacks, and that our data doesn’t show that AIGC offers what could be considered an unfair advantage is the main reasons I have not really felt that we need to ban AIGC from the jam.

It’s Not That Hard, Just Do It Yourself!

Ok, so in the interest of being up front, I really hate this argument. While I understand the intent, this argument overly simplifies a really complex topic, and personally I think is used more often in bad faith than not. Generally speaking I have 3 major issues with this argument.

First, this argument tries to reduce everything to the fact that many of the skills a person may use AIGC have a fairly low skill floor, but in doing so ignores the fact that this is a much more complicated issue than on the surface. While many of those skills may have low skill floors, they also have very high skill ceilings and a variety of skill curves that can make learning difficult beyond the basics. Also, within the context of making games the minimum skill required for a game varies by the requirements of the game and almost never aligns with the skill floor.

Using programming as an example (since that is what I know personally) if you count very basic web dev programming can have a very low skill ceiling, but for most games this will not work. At minimum you would need at least some app dev experience which is well above basic web dev. Also, the skill curve for programming tends to be a long plateau with sudden cliffs which makes it very easy for inexperienced devs to think they know more than they do and get into trouble without a more experienced hand guiding them. At the end of the day everyone one of these skills, from writing to art, may appear deceptively simple but the realities especially within the context of games are much more complex than a simple “get good” argument.

Second, I feel this argument is abused in bad faith (intentionally or not is hard to know) to either discredit the opposition as lazy or make the opposition feel stupid for “not getting it” or both. The reason I feel this way is it is very rare to hear this argument followed with “and I would be happy to help you learn.” I very rarely see or hear about anyone leveraging this argument then following it up with a sincere offer to help the person learn so they can better themselves. It’s more often I see it posited straight or followed up with a “I did it so you can too.” That last part is particularly irritating to me as it is the equivalent of me saying “you should stop using RPG maker and learn some real programings, since I self taught myself it in junior high using a book so old it assumes you don’t have a computer monitor.” It trivializes the amount of work you actually had to put in while simultaneously being unhelpful.

I try to walk the talk with the advice as well, so I am not simply just complaining about the fact that I don’t see people offering to teach or help. I have never used this argument specifically myself, but whenever I see someone asking about using LLMs for programming, I explain the downsides and traps of using it especially without some programming knowledge, but then I offer to teach them if they want to learn. I am honest with them, I tell them it will take some work, but I would be happy to help them learn it if they are interested. Not everyone takes my offer, but, like I said above, if the decision is informed it is their call at the end of the day.

Finally, probably my largest issue is, intentionally or not, this argument belittles the effort and time of those who worked hard to learn that skill. After all, if it is really that easy then it should have taken little to no effort at all to learn, right? By underselling the difficulty and complexities of those same skills that the argument is being used to “protect” you trivialize it and I feel disparages those who put in the time and effort.

But Isn’t the Entire Community Against AIGC?

Honestly, I thought about the same at first. At first glance with how vocal anti-ai sentiment is one would think that is for sure the case but as I started asking around I found the situation was more complex then I originally thought.

Now I want to be clear, I think overall sentiment still leans more anti-ai, but after my discussions with artists, writers, devs, and random users it is a lot more split then what was originally believed.

What Did I Learn from Artists?

I do want to say that while most of what I say here should always be taken with a grain of salt, artists were the smallest group I was able to talk with so I am less confident about how representative what I learned is. That being said I still think it is worthwhile to share.

This one surprised me quite a bit personally. With how much anti-ai sentiment is targeted towards image generators I assumed Artists would be near universally against AIGC. What I learned though is the artists I talked to had more mixed opinions then I originally thought. Even within friend groups, some were indifferent to the use of AIGC where others were fully against it. Even more surprising, while they had distaste for how the models were trained, expressed interest in discussing or even experimenting with how it could be used beneficially for artists.

It seems like behind the scenes while the opinion does lean negative it’s more mixed then I expected. It also seems to depend a bit on how confident an artist is with their own work with artists that told me they were confident in the quality of their work generally holding more neutral opinions. A common thing I heard though is many of the artists I talked to did say they felt there was a good chance that they would be harassed or run out of the community if they expressed any opinion other than full anti-ai support so preferred to stay quiet or tow the line.

What Was I Told by Writers?

Writers were by far one of the more interesting, and saddening, groups to talk to. Generally speaking, writers seem to be the most neutral out of the groups I talked to. Many admitted they use LLMs to help them with their writing in one way or another. Usually it’s about what you expect, with many using them to workshop ideas, deal with writers’ block, and even as makeshift editors to proofread and suggest alterations to their story. Though a few said they use them to impersonate their characters to help them nail down the voice they want to use for them before even writing the story, which I thought was really clever. None said they used AIGC directly, only to assist them.

When I asked them why it seems like they chose to embrace the tech, even reluctantly, their answer made me quite a bit sad. They basically said it was adapt or die, and they didn’t think anyone was going to come save them, much less stand up for them. Many I talked to expressed that when LLMs were first opened to the public they felt a lot of people did not take the threat to them seriously and even a few were optimistic about what it could do for them, and that it really didn’t start to be perceived as a threat until image generators started to increase in quality. Still though the focus was on the image generators and not on writers, so they had to figure out how to use the tech to ensure it didn’t replace them and other ways to fight back on their own.

Even sadder was that a small but fairly large group of the writers I talked to expressed lasting resentment against artists and their communities over the incident. They talked to me about how they always felt like their artform was always treated as lesser, but were largely ok with it because they knew that images were flashy and naturally drew attention. But, when not very many people stood up for them, but then they were told they had to stand up for artists it left a bad taste in their mouths. They stated that it felt like artists were demanding special treatment when they had to suffer, and even with all the anti-ai sentiment floating around right now they still feel like writers are still an after thought at best or being used as a tool to protect artists at worst.

Personally, I can see why they feel that way. At work we have a joke that good design doesn’t sell leadership on an idea, colored circles do. Images draw attention better than text and require less effort to enjoy, even if a well written story may be better than any one individual image. Also, writing has always felt like a second class citizen in most art circles to me. The written word is usually an afterthought on most gallery sites, and when I compared text adventures vs VNs with our own site’s data there was a huge gap between the two with pageview, likes, and posts on each topic.

I have even seen it when discussing AIGC with other people and even some artists. A common response when I asked about writing was either a complete or partial dismissal of it, or a saying that writing is so easy that AIGC shouldn’t be used anyway. I do want to be clear though quite a few of the artists I talked to did show respect for writing as an artform, but even they admitted that there has always been a bit of elitism against them mainly perpetuated by an artists fanbase and on occasions by other artists themselves; so they understand why some writers may be feeling like they were abandoned and resentful about the current situation.

What Did Devs Seem to Think in General?

Generally speaking most devs viewed the use of AIGC more practically. Many disagree with its use but don’t want to outright dismiss the tech and felt that if a dev felt they needed to use it their decision should be respected. But, many also acknowledge that most devs that use AIGC at least straight shouldn’t really try to sell or sell access to their games just out of respect for the fact that how these models were trained is very morally grey at best, but do admit it is hard for them to not see the justification of charging a price that the dev feels is fair for their work alone.

How do Users in General Seem To Feel?

From talking and looking over our site data, most general users tend to be very apathetic to the use of AIGC. They are mainly there to look for games and/or support specific creators/projects they tend to like. A fair number we did talk to though did express annoyance at how frequently and aggressively games that made use of AIGC were attacked; though reasoning varied from a general distaste for witch hunts to the incidents drawing attention away from games they thought deserved it more.

What Does This All Mean Overall?

As I said above what this all leads me to believe is that overall community sentiment is a lot more mixed than anyone thought. The illusion of this “unified front” against AI was more perpetuated by the fact that the anti-ai side tends to be so vocal and aggressive in imposing their views this suppressed many opinions to the contrary. This has led to an echo-chamber effect where any dissenting opinions are suppressed (either intentionally or due to fear of being called out in general) giving the illusion that anti-ai support is actually greater than it actually may be.

This alone is one of the main reasons I really don’t want to take a hard stance one way or another. On Weight Gaming alone we have seen our fair share of dog piling and witch hunts against anyone who even talks about AI. I have seen quite a few discords where AI talk is not banned but anyone with a pro-ai or even wanting to discuss if the tech could be used to benefit and not harm artists are instantly booted from the server while allowing people to smack talk AI all they wish.

At the end of the day what does this sort of behavior really help? Many of these incidents I have seen were from people who were exploring or wanting to discuss the tech, not some hard core ai-bro looking to cause trouble. So after that what do you think that person is going to do? Repent? Why should they when they have already been banned and told they are a bad person. Look for an artist to work with? They were told no artist will ever work with them. So what will they do? This is how you push people further into embracing AI and breed resentment overall.

Why Do You Think Devs Turn to AIGC?

Quite a few of the more vocal people on the anti-ai side will say devs turn to AIGC because they are either lazy, dont care about creative integrity, or are trying to fuck over some one (either artists or the community) in some way. The reality of it though is far from those accusations in most cases. For the sake of my sanity and to try to cut down on this monster I will only be focusing mainly on visual assets though what I am going to say can be generally applied to writing, programming, sfx, music, ect though there is some more complex reasoning for a few of those then what I will be outlining here.

Generally speaking most devs I have talked to say AIGC was not their first choice. Instead it was a final option when they either failed to find someone to help them or thought they could not afford someone to help them. And if we are being fair this makes sense on the face of it, especially for visible art which we see AIGC most commonly used for.

If we are being honest wg and expansion fetish games are very unique in their requirements, especially for an indie scene. One of the most obvious ones is that these games are much more asset heavy than standard games. To give an example lets look at a JRPG quickly. Lets say you want to make one overworld sprite with a standard 3 frame walk cycle. For a standard JRPG that would be 3 separate directions (you can flip E/W to save work) at 3 images for each directory. This is 33 which gives a total of 9 images an artist would have to draw. Now let’s say you want to have 5 weight stages, now its 53*3 for a massive 45 images which, obviously, is a 5x increase in workload. Now let’s say you want a standard 4 person party. Now you are looking at 180 images just for that where it would be ~36 images with a standard JRPG.

This can make working on games daunting for many artists. Even when offered payment that goes beyond their usual commission rates, the amount of work is a very common reason I have heard for an artist not wanting to join a project. Further compounding it, it is not feasible to simply throw more bodies at the problem like in B or AA studios. While I think fetish games can be commercially viable the max size team that I think could be supported is maybe 5 people, so most groups would be lucky to have more than one artist.

This means in general for most teams finding an artist willing to work under that load or even being able to pay them a fair wage is a very tall order and that is before accounting for other assets the artist may have to produce. Usually, asset packs can be used to help reduce this workload but there are no good asset stores for fetish assets. And while some artists offer character or template packs they are few and far between and usually have restrictions that wouldn’t make them viable to be used in a game project.

Now there are some people who would say it’s better to have no visual art at all then to use AIGC, but I think they are either blind to or choose to ignore the overwhelming advantage that visual art does lend to getting a project noticed and engaged with. Comparing pageview, likes, and posts of projects tagged with VNs vs those tagged as Text Adventures shows a very sharp disparity. On average VNs get 107% more page views, 249% more likes, and 136% more posts on a topic then Text Adventures tend to.

Now there is an argument that due to the fact Text Adventures have such a low barrier to entry this could be a result of a lot of low grade Text Adventures and VNs tend to be higher quality on average due to the higher barrier to entry. This could be true, but let’s look at the text adventures that are marked as using AI. When compared to the same VNs, Text Adventures that use AIGC fare much better. The 170% gap narrows to just 17%, 249% becomes 104%, and 136% actually becomes -3% (though this may be more due to the fact AI fights that tend to break out in those topics, skewing the numbers). Also, when you compare it to itself, Text Adventures that use AIGC generally perform ~50% better on average.

I point this out not to try to support the use of AIGC, but because I think there are quite a few people who are eager to ignore this or unwilling to even consider that a dev may have a reason for using AIGC beyond malice. There are many people who I have talked to on the anti-ai side that are quick to label anyone who uses AIGC as evil and undesirables that need to be purged from the community. It is not an uncommon response when I bring up a dev leaving the community due to harassment to hear “good, they were just here to harm the community anyway” with no justification or reasoning beyond they used AIGC.

In polarizing topics like this it can be very easy to dehumanize each other, but this makes us blind to the roots of the issue. Everyone wants their game to be noticed and feel like their work has been worth the effort, and claiming the motivations of all of those who choose to use AIGC as being lazy or malicious is ignorant at best and self-serving at worst.

So with all this in mind the question I wish to ask all of you, is it really a surprise that some devs do choose to turn to AIGC since there are not really any viable alternatives if an artist falls through?

Continued below

30 Likes
2

What are We Going To Do?

As I have said on several occasions, I am personally against either an outright ban for or against AI. I feel there is still a benefit to being able to openly discuss and explore the tech, but I can not deny that it has been a moderation nightmare. While I am against taking a more hardline stance I will not be taking it off the table. What I would like to try first though is see if bringing the hammer down might help without forcing us to fully ban one side or the other. It might be a hard balance to strike but I do see people have civil discussion about AIGC here on the forums and I do not wish to punish those due to the actions of a handful of people that are causing issues.

While I always prided ourselves on trying to take a hands off approach to moderation and attempting to encourage people to self correct before taking action, I think this stance has proven to be detrimental to us when it comes to AI since it slows our response time in general and some people do try to abuse our system to try to delay action against them. Due to this I plan on working with Krod about taking the gloves off for our moderation team. I’ll make a post detailing what we plan on doing once it is sorted out but expect to see our mod team being allowed to be a lot more strict on unruly behavior and what could be considered harassment of users. As well as being allowed to hand out punishments like account silencing and temp bans which used to be reserved for more disruptive behavior or would not be given out in the past without at least one warning. I do hate bringing the hammer down like this, but the situation has been out of control for too long and we do need to put our foot down.

Beyond that I still think, as stated above, that an asset store can do a lot by providing an alternative to using AIGC. I will be talking more about it in the final post, but I am still dedicated to trying to provide that asset store to the community. It is my hope that by providing one we will give a viable alternative to AIGC to devs, but will also allow those who are harmed by AIGC a route to make at least some supplemental income.

Final Thoughts

With how much I focused on the anti-ai side I am sure a few people would be thinking I am just defending AI. The fact is the blatant abuse I have seen of AI mainly by companies has really soured me on it in general, and it’s very fair for people to express their general distaste for it.

My issue is I hate seeing how everyone has been fighting each other over this.The anit-ai side feels quick to villainize while the pro-ai side feels quick to dig in or want to hit back. And I know I was harping on the anti-ai side but honestly if I get a message about something bad happening 9 times out of 10 it’s unfortunately started by someone who hates AI.

And I wish I could see even a quarter of all that energy and effort put toward pressuring our politicians to strengthen IP rights for smaller creators or consider regulation on AI companies. Because frankly, arguing such a point on a fat fetish site does nothing to actually help the situation. What we can do as a community though is not be so quick to villainize, try to understand the root reason people may turn to AI, and figure out ways we can deal with those root causes together.

Now I am never so proud to ever think I am 100% right on anything so I welcome anyone who wishes to make their own points or debate mine to feel free to do so civilly below with a special emphasis on civilly. As I explained above I am no longer going to handle this with as much of a soft touch like we have in the past, so don’t expect any warning from Krod and I and we will not hesitate to silence accounts as soon as we feel a user is getting out of hand.

29 Likes
3

I’m glad you’ve made this post. I agree with your viewpoint and share your lived experience about the amount of vitriol this topic has generated in our forums. I hope those with a strong anti-AI stance read your post to completion; although it is long, it is clear you are not just “pro-AI”.

In my opinion, you are right - blanket-banning AI does not put the genie back in the bottle. An undemocratized AI sounds worse, not better. Taking it out on the little guy only hurts the little guy.

I hope that, in the far future, this issue can be relitigated and apt compensation can be given to every person that’s been stolen from, if that’s even possible on the sheer scale of tech companies’ thievery. I share the fears of those against AI, but let’s not take our frustration with the world’s elite out on some guy in our tax bracket. Like you mentioned, the way a few community members have been treated here gives elitism, not unlike how people have looked down on text adventures, RPG Maker games, and even the traditional media vs. digital art debate that happened when digital artists started to become more popular. I hope we can be nicer to each other.

5 Likes
4

it really is not a easy call on the one hand it really is like a co pilot for bringing ideas to life while also having the ability to destroy things

1 Like
5

Good post. I especially agree that the hard-lining has become an issue that has turned more people away from the anti-AI side. This is part of why my stance on AI is “depends on who I’m debating/arguing with”.

From a writing perspective, I wholeheartedly agree with the points you’ve brought up. I avoid using LLM in any of my writing/dev work to avoid the whole Slippery Slope thing, though I can confidently say that AI-generated text content absolutely wipes the floor with me (to the point that I’ve admittedly fallen victim to the distraction that is AI chatbots). From a Dev standpoint, while I trust LLM-written code about as far as I can throw the server it was compiled on, I see no issue with using an LLM to learn how to write certain types of code. After all, most of my programming knowledge was found on Stack Exchange and other forums.

Visual media? This is where my stance really goes to “depends who I’m arguing with.”

I am not a rich person. In fact, I could be called poor. There is no way I could afford to hire an artist for one of my projects (I consider it a blessing that Bun was willing to draw up thumbnails for my Lipomancer games). And given that I have no intention to sell any games (and the itch payment processor situation has made it perfectly clear that I am not to allow donations), I don’t think I’ll really be able to generate enough money to hire an artist.

To that end, I’ve only got three options for having visuals in my games: Don’t have visuals, use AIGC, or try and convince an artist to put up with my shoddy communication and work for exposure (I’m not gonna do this one, so we can strike it from the list right away). That or spend a long time that I could be spending writing/programming a game on building up my drawing talents enough to meet my own standards.

For the time being, I’ve chosen the first option. While it’d be egotistical of me to say it’s the sole reason for my lack of popularity (I’m sure there’s plenty of problems with my game aside from a lack of visuals, plus I’m a strong believer in the theater of mind), I can’t deny that the statistics that Grot have brought up are probably not untrue. I can hardly blame other Text Game devs for choosing the second option, especially when they at least put in the effort to curate their AIGC.

I will not make any comment on the hardline Anti-AI folks, as anything I would have to say would be rehashing what Grot has already said (as if what I’ve said up until now hasn’t already been doing that lol).

EDIT: I spent so long writing this response that I forgot I already covered the last part of this wordblock in the first part of this wordblock lmao

9 Likes
6

Jeez I wrote a lot about this.

TLDR: please think about how genAI models are created and sold, sometimes things are bad even if they’re convenient, consider the implications of your arguments, site admins need to make an actual decision about all this.

This is like nine paragraphs I'd like to touch on one point real quick before getting into the meat of it, mainly this that you said early on:

You (you as in the site administration as a whole, not just you specifically Grot) are in a leadership position in the community whether you like it or not. You all make the rules and run the forum that everyone here uses to share their work and ideas with each other, you outline how conversations should and should not be conducted, and you provide the general direction this site heads. Abdicating this leadership means a lot of people have questions that aren’t getting answered, people don’t know what is and isn’t allowed/encouraged, and leads to the big thread of complaints that prompted this latest set of responses. A hands off approach doesn’t really work in the case of site moderation, because it just means people are going to be jerks to each other (purposely or not) because no one told them no. Plus having (even internal) site-wide stances means you all have to talk about these things before they boil over into a whole shitfit.

On to the genAI issue. My main ethical stance comes in two parts. The first part is that if anyone’s work is an essential part of the creation of their products, i.e. the training data, then any non-open source work used without compensation is theft of labor. If I made a program that generates 3D models of cars but it just mixes and matches a bunch parts of other people’s car models and sold that program, I would be stealing their work. If something is making a profit, the people that allow the profit to be made should either be compensated or actively volunteer their time with the knowledge they won’t be compensated.

The second part is stopping a bad standard before it becomes a standard. “AI is inevitable” is something I hear thrown around a lot, and I’m not going to dispute the technology itself isn’t going away. But just going “Welp, the companies already got away with scraping all that data and this is super easy, I guess I have no choice but to use it.” is a really short-sighted and kinda selfish mindset. They’re getting away with it because people don’t really care that it’s built off of other people’s work. The only way something becomes a standard is if enough people accept it. The more people that say “No, I’m not using it until it’s done ethically”, the more incentive there is to actually do it without screwing over the people whose work its built on. And this is still a pretty new industry! There’s still time to curtail these practices before they become a standard, to not make it a regular part of life or an integral part of any workflow! The tech may be inevitable, how it’s currently implemented is not.

This is a bit of a simplification, but I’ve really only seen one pro-AI argument, and that’s the argument of convenience. Especially for (and I’ll touch on people charging in any part for genAI content) people who do this as a hobby. It’s convenient to not have to draw, or write, or code, or make music, or learn how to do any of those things, or to gather people who can. I get it, I’ve had to wear all those hats. It’s slow and it’s frustrating and you’ll never be able to work as fast as people who use the convenient tool. It’s very convenient, and you have to decide what you’re willing to care (or not care) about for that convenience. Tracing is also convenient. So is just using someone else’s art you found online. Using a program made through “using someone else’s art you found online” process times a million makes it a lot easier to argue you have no personal responsibility, but the “someone else” still gets their work used either way.
At least using stock art or asset packs you’re paying the person who made the art. Hell at least if you’re just straight up ripping content wholesale you can think about the difference between taking sprites from a 20 year old Nintendo game or Jimmy Commission who needs to make rent.

As for charging for projects with genAI content, I don’t actually have any new arguments. It’s the same thing but more skeevy since now instead of the genAI company being the only one profiting off of unpaid work you are as well.

As for everyone who feels like their project won’t get attention without visual art, I don’t really have a good solution for that unfortunately. It’s the same thing that makes silly meme posts a thousand times more popular than stuff with way more effort put into it. I still think you shouldn’t use genAI for attention, and I think it’s counterproductive in the long run. Sure you get a few more eyes on your project (silly memes could probably do that too), but how many of them care about what you did vs the AI pics that anyone without your skills could have generated? If they’re just placeholders and you want to attract the attention of an artist to replace them, is anyone going to be able to keep up with the output of genAI? Are you going to be able to adjust to working with an artist when the alternative is just so much more convenient? Is your willingness to resort to the convenient option going to affect how others view your willingness to collaborate? The path of least resistance is hard to leave.

Quick tangent, I would also just add a word of caution for anyone relying on the tech for anything important, the genAI boom is a pretty clear case of blitz investment. There’s been a ton of money poured into this on the expectation of paradigm shift and a massive return. Once the initial investment funds run out the hope is that the tech will be so firmly entrenched that people will have no choice but to deal with the inevitable price hike for any of these genAI services.

To make sure I’m touching the main points brought up, if you are anti-AI for pete’s sake get a proper reason for your stance. I’m tired of hearing all these terrible generic arguments “oh you have to do all the work yourself or you’re lazy” (asset packs or just having a team), “machines can’t make art” (so ill defined as to be meaningless), “technology is putting people out of business” (broad issue with technological advancement in general, not really an AI issue specifically, at the very least read up about luddism before saying this), “it’s bad for the environment” (yeah welcome to most industries, also not an AI issue specifically). Choose an issue that is unique to the current implementation of this technology and think about it for more than two seconds so you don’t end up repeating mantras that don’t actually hold up.

All in all I feel like your final thoughts are still pretty non-committal and trying to appease all sides of the issue rather than creating an actual site policy that will stop these endless debates. If you want to see everyone stop fighting about this on this site, make a rule that stops the fighting beyond the “stay civil” that hasn’t worked so far.
It’s pretty clear from what you’ve said that your only main sentiments against the use of genAI is from companies and not individuals, so I doubt there’s going to be any site policy against the use of genAI. Just make it official and either ban the radical anti-AI fringe or ban anti-AI discussion so we can cement the site’s stance, staff can shut down arguments by citing it, and we can all move on from this.

20 Likes
7

So, there’s obviously quite alot to be said on this topic, as Grot has already demonstrated quite excellently. I agree with much of what’s already been stated above, but I want to provide my view because I have quite a bit of personal experience in regard to this topic.

First, I want to acknowledge the public-facing position that I have in regards to AI - For better or worse, I’ve become seen as someone who represents the pro-AI side of the argument, as I’ve released a number of games that make use of AI tools. Wasteline, the jam game that sparked one of the more significant AI feuds on the forums, made use of AIGC due to decisions that I made when it was being developed (and solely me, the others who worked on it had no part in the decision).

I want to make something extremely clear, though: I do not like AI. I think that anyone who actually says that they “like” AI is either ignorant to the effects that it has and will have on creative fields, or simply lacks empathy for the people who have dedicated so much to their respective crafts and are now being threatened, and thus I typically find it hard to respect a stance like that.

However, I don’t believe it’s a very common opinion to actually think that AI is largely positive. You would be hard pressed to argue against the fact that a large swath of corporations have begun culling many of their employees so that they can replace them with inferior, unpaid machines, and any person with at least a shred of humanity can recognize how shitty that is.

As someone who spends a hell of alot of time writing and coding, does it make me feel good that an LLM can spit out something of at least somewhat comparable quality when it would take me hours to do the same? No, obviously not. There are certainly arguments to be made that AI is a good tool for the average person to get help in areas where they lack expertise, but I still think that it would be a net positive if generative AI didn’t exist.

So, as someone who has a relatively negative view on AI in general, why would I use it? Well, because it’s hard to deny the uses that it DOES have for the average person, and, of course, the average game developer. As Grot said, games inherently make use of several different types of media. As a solo developer, possessing a skillset that can successfully cover all of these different fields is a practically impossible ask, and thus generative AI’s ability to fill in for basically any of them is an incredible asset.

Before I continue on this point, however, I want to preface it with something else. “Solo developer” is the key here - In my opinion, someone who works on games as only a hobby and doesn’t have the financial backing to hire help in areas where they’re lacking is perfectly valid for wanting to supplement their work with AI tools. However, when a developer has the resources at their disposal to actually bring on more people to support a project, using AI to cut costs isn’t nearly as excusable.

I’ve been successful enough with my games to now be able to count myself as the latter case, and thus I’m happy to say that I’ve started to move away from any and all AI tools. My intention isn’t to advertise myself here, but I just want to prove that I’ve put my money where my mouth is on my AI stance. The current project that I’m working on at the moment is a remake of Wasteline, the aforementioned jam game that caused a bit of a stir with its use of AI, with all of its AI-generated visuals replaced with real art. As I said, I think that those who are financially comfortable enough to afford human-made assets don’t have any good reason to make use of AI, and I’ve been trying to hold myself to this standard.

Anyways, with that tangent over, I’d like to now explain the use case where I think AIGC has at least some merit. When it comes to this particular space, I think that it’s for the best when people are given more tools to develop games. The forum is already pretty niche, and the amount of at least somewhat high effort projects that exist here isn’t exactly staggering. Due to the general lack of games in this space, I feel that we can always use more people pitching in to make stuff to cater to our interests here, and in my opinion AI tools do nothing but enable the process.

Of course, lowering the barrier to entry will always lead to more slop. There are plenty of AI-ridden, low-effort projects and dime-a-dozen chatbots that we’ve all seen pop up, but it doesn’t take more 10 seconds to just pass right by them. However, in my opinion, the amount of games that have been released that make use of AI tools but then excel in other areas is absolutely worth the occasional bit of annoyance that the lower effort stuff causes.

Naturally, this is only my opinion. I know well enough by now that there are plenty of people who would vehemently disagree, but I’d like to provide one last personal anecdote to hopefully show why AI can be a net positive in this kind of space.

I can say for a fact that I would not have gotten into game development if not for AI. One day a few years ago I decided on a whim, “hey, I could try making a game.” Unfortunately for me, though, I didn’t have either the artistic talent nor the financial means to realize the visual element that I wanted for my game. I knew that most people generally wrote off games that didn’t have any visuals, so I felt a little hopeless at the prospect of getting anyone to notice what I wanted to make if I wasn’t able to provide any art.

Thanks to the fact that image generation technology had just kicked begun to kick off around then, though, I felt like I had been given the opportunity to actually take a crack at what I had wanted to make. So… that’s what I did. I made a game with AI-generated visuals, and then I made another (much higher effort) one after that. Sure, I put a ton of work into the writing and code that went into these projects, but I really doubt I would’ve had either the motivation or success that I eventually found if not for the fact that a broke hobbyist gamedev like me had been given a tool to much more easily complete a significant piece of the game development process.

Fast forward to now, I’m still working on games (now without AI), and I still really love doing it. It’s one of the biggest things that drives me in life, and I doubt I would’ve found this pursuit if not for the fact that my creativity had been fully enabled by AI. I don’t really love the fact that I owe a big part of my success to ethically questionable generative software, but it’s impossible for me to deny that it’s led me to what I’m passionate about, and I’m glad to be able to say that I’ve begun to move beyond it towards greater things.

So, for all the opponents of AI out there, I know that you probably hate much of the work that I’ve done. At the very least, though, let me say that we have more in common than you may think. We both want what’s best for the forums and what’s best for the development of these niche fetish games, and I can fairly confidently say that neither of us really like the greater implications that AI has for society. I just hope that my essay here has helped to convince you and everyone else at least a little bit that this kind of thing can have positive effects on the community that we’ve built here.

13 Likes
8

Fantastically written and (especially in context) reasoned post @grotlover2

Speaking for myself, as a dev that makes use of AIGC I have a few things I’d like to say that I’ve not really seen said by others. This is not to try to “convince” anyone of anything. Just sharing my experience.

  1. Even using AIGC I put a lot of time and effort into making my VN. And by that I mean every element. Those that I hand craft myself and that I use AIGC for. Not one single piece is quickly knocked out as “good enough”. True when it comes to visuals and audio I have sometimes got lucky and got what I wanted from the generators first try. The majority of the time though it takes hours of work refining and regenerating. Even once I have the raw outputs it doesn’t stop there - I do post production work on the assets. Sometimes that’s simply removing the background in an image, or tweaking the bass up and cutting out noise. Other times its pixel-by-pixel image alteration. And admittedly, sometimes after spending hours trying to refine an image I still can’t get what I want and have to “settle”. I’m a practical man - I understand the limits of the tools I’m using. To me - while the visuals help sell the story, they are ultimately secondary so I believe it is justified to have the occasional “bad” image.

  2. I mentioned before how I can get lucky with my images and audio. That “luck” is built upon a lot of work and experimentation learning how to get the model to produce something close to what I want. I see a lot of AIGC being referred to as slop and to an extent I would agree with that statement - bad in gives bad out. In that regard the use of AIGC is a skill like any other in my view.

  3. I see so many VNs that make use of AIGC that all look exactly the same, like every one is using the same asset pack of characters with palette swaps and using them a little differently like its 1992 and its Scorpion vs Sub Zero all over again. I know logically that this is down to using the same widely available model and a “popular” art style. But it is a bit depressing seeing how few seem to want to deviate from that.

  4. While I can’t speak to how the training data was obtained for the image model I use I try to minimise the impact my use of it has. I don’t run it on a GPU farm in the cloud. I run it locally. It might be a drop in the ocean - but I am not blind to the harmful side of AIGC so try to limit any “harm” I might cause in using it. I agree that a lot of big tech companies are unethical in their practices. If I could boycott them entirely I would, but the way things seem to be going these days AI is getting so integrated with everything its becoming difficult to avoid in every day use.

There’s more I could say, but it seems I’ve run out of time. Might come back and edit later for a couple more points. In closing summary, I am generally against the unethical use and practices involved with AI and AIGC. That being said, I believe if used responsibly its a valuable tool to plug the gaps - to help bring a story to life but shouldn’t become the story itself. Game design is a difficult, intense process whether you make use of AIGC or not. As far as I know not one single one of us creating games here is an industry professional. We are hobbyists and enthusiasts. Ultimately - we are all here for one thing: To enjoy playing games that cater to our niche interests. In that spirit in the words of Rodney King “Can’t we all just get along?”

9

I will try not to butt in too much as im not a dev and for art, i have only done a few sketches or quick stories for myself over my life so my opinion might not have as much value. I hope it doesnt seem as changing topic or attacking but you yourself posted recently a game, one that i loved to play and uses different characters from different artists with different art for each. On the art side of it, wouldnt it have made it pointless if it all had been made on AIGC, since it would only need a single person to make instead of a collab of different souls working together. I dont know everyone who worked on it, so maybe some used AI and i cant see it as its harder and harder to notice or could even have been in the code or writing as those are much more difficult to notice but there is something there unique.

I also understand that people in such a niche thing like wg arent looking for uniqueness all the time and want something that goes for the point simple and quick, that dont care about mechanics or plot elements and just want number go up or button goes pop! (i myself have felt like that at times). I have supported artists that have used AIGC backgrounds to make their works come quicker and have slowly worked on their own official thing for the side but in the past they probably would have used just a stock photo online as many did before it was a thing. When there wasnt AIGC, people found ways to cut corners, its just this is probably the biggest tool of them all to use that and is starting to get noticed by either big companies using in the products like games, movies and so on or probably when we receive an email. Its impossible to be rid of it now as you have said but i do believe it needs some moderation and im hoping for the people in charge of such moderation to be on the lookout for.

10

Personally, I was practically waiting for this post. Your explanation of the whole situation and your thoughts are clear. Frankly, I was surprised by the artists, developers, and writers (writing an interesting story is not easy; I’ve tried it myself). Unfortunately, people tend to use negative emotions for destruction rather than creation. While I don’t support the LLM, I can’t deny that it can be used for good, not just harm (though I can still draw; writing music would be more difficult, but solvable, and programming is my biggest challenge). I’ll most likely start using the LLM to learn programming, but I’ll also try to ask people to at least share their understanding of how it works (I don’t speak English, so programming can be difficult, but I believe it’s necessary for independence).

I won’t insist on my opinion, which is slightly negative towards AI, unless I’m pressed and someone writes or says something like, "Are you still making games with your meat fingers? “Wake up, you can do everything you do with a couple of clicks now, idiot.”

The last thing I can say about my personal opinion on LLM is that the game I create without LLM is mine, MINE. It’s important to me, at least in this way, to assert myself and say that I’ve achieved something on my own. As for everyone else who uses AIGC or something I don’t like, I’ll just leave them alone, saying that’s their business, and what I do is mine.

5 Likes
11

I don’t know, I just feel like having stricter moderation isn’t enough, I guess I was expecting a more impactful solution. It just feels non-committal, as if you’re trying not to step on anybody’s toes, instead of just putting your foot down and making a more substantial decision.

I’ve noticed that a lot of the issues discussed over the past month, was allowed to fester and become an actual problem due to not taking action, and not doing something about it when you needed to. That’s why I’m concerned about not enough being done to deal with this issue specifically, you’ve made a lot of great points and I get wanting to be cautious when dealing with this situation, but I’m not seeing much actually being done about the situation.

There needs to be a more clear and strict rule other than just “stay civil”; something that moderators can actually cite and use to moderate more efficiently, instead of trying to decide if someone is being uncivil or not constantly. AIGC isn’t gonna get banned obviously, judging off of everything you’ve said so far. The issue currently seems to be more about the anti-ai extremists and these AI discussions turning into war zones, resulting in people getting hurt and/or leaving. You either need to ban the anti-AI extremists causing this shit, or if everyone is just incapable of discussing this topic without attacking and hurting each other, ban anti-ai discussion all together. I just don’t think the same ol’ moderation, but with a little more backbone put into it, is gonna cut it.

I am actually really interested in your idea of an asset store though, and I’m excited to hear more about it. I have had thoughts before of creating some furry WG assets, to hopefully help deter some people from GenAI, and give people in the community a viable alternative to GenAI. It’d be more than happy to contribute to it, if you end up going through on it!

10 Likes
12

I personally grew sick of the AI vs human art conflict because it was always planned to be a pointless and distractionary argument that pitted people who aren’t even artists in their own arguments against people who… aren’t artists and acknowledge such

instead of the discussion being about how these companies FOR NEARLY 2 DECADES have been talking about AI as they envisioned it being used to do exactly what it’s doing and pushed all development of AI in that direction, artists having it too good in scarcity aspects and too bad in the tools to increase productivity aspect, or how to make sure AI would be used on an individualist level… people took the bait and made it a morality argument on how much labor must be put in for it to be defined as “art” (an argument so old that it can be traced back to the invention of photography)

this inevitably means most anti-ai art arguments are made in such a way that people hilariously dismiss themselves as artists such as:

  1. digital art creators who rely on simplified styles such as shadeless styles
  2. editors of digital or real life media
  3. people who use pre built assets to create static scenes in a 2d/3d sense (think kisekae, sfm, mmd, Koikatsu!, ect)
  4. software based animators
  5. visionaries trying to bring an image to life without having the personal skills to create everything
  6. many other roles that dont involve spending 30 hours sitting at a canvas waiting for the previous coat of paint to dry

it’s even worse now since in the face of the inability to have their way the anti-ai groups are doing what any group who cant get a moral hegemon over an argument do when their morality scale isn’t accepted by anyone outside of their extreme, they turn to underhanded tactics and blatant “violence” (metephorical in this sense) against anyone less than willing to hang their heretics

3 Likes
13

I want to highlight these points in particular to illustrate something that’s a pretty big disconnect between folks going “I have moral concerns with the use of genAI” but then turning around and saying it’s fine to use it in specific circumstances. The problem here is that the “good reason” for using genAI is “I want to”.

“I want to make a game”

You can make a game without AI.

“I want to make a game with visuals”

Well draw up some assets then.

“I want to make a game with visuals beyond my skill level”

I suppose you could develop your skills over time to improve your skill level

“I want to make a game with visuals beyond my skill level right now”

I guess stock assets or working with someone else is an option.

“I want to make a game with visuals beyond my skill level right now without paying anything or networking enough to do a collaboration”

Well shit, I guess if you keep adding conditions on the thing you want to do it does narrow down your options to either not doing it or using genAI regardless of your ethical hangups.

The problem with this argument is that someone without other resources and someone with other resources are still using the exact same programs and paying the exact same money to the exact same companies. Saying you’re going to be ethical about it now, when not using it isn’t the difference between whether you make a game at all, just means you’re willing to compromise as soon as it’s convenient for you. Sometimes you have to make a choice to give up on something you want for moral reasons. And it’s okay to change your stance on something moving forward, even if that means admitting you did something wrong that you’re not proud of. Otherwise you’ll just end up trying to justify your past actions by going “but I really really wanted to so it’s okay”.

18 Likes
14

I can’t help but notice you only referred to visual AIGC, not text or code AIGC.

Edit: double post for some reason, weird.

2 Likes
16

That’s because it was a reply to someone talking about their personal use of genAI to make visual art. The same principles still apply to any use, just replace “visuals” with “writing” or “code”. I do think the use of genAI for both code and writing are also bad, the fact that there was an entire chatbot category on this site still leaves a bad taste in my mouth frankly.
I will say that I think most people are going to be arguing about the visual side of things unfortunately, but that’s mostly due to people feeling a lot more compelled to add visuals to their work for popularity reasons.

4 Likes
17

I think one source of tension is that the general lack of curation on weight gaming is more of a problem in the age of AI.

It used to be the natural skill barrier required to write, draw, or program something yourself meant that by the time someone had something to submit to the site, they’d put in some time learning and developed some minimum level of discernment. So even without any official curation, it used to be that the average project was at least worth taking a peek at.

4 Likes
18

This comment in particular is a point that people who have an evident dislike of AI seem to always have as a secondary reason for their feelings, rather than their primary reason. I do agree with your sentiment, because I think it cuts to the real issue of why AI would leave a negative impression on most people.

AIGC, for the most part is something that most people for better or worse do not care about in terms of morality, and never will. While they may feel some level of distaste in the theoretical of taking away work from someone with creative ability, or the more nebulous concept of environmental degradation, the truth is that most people are only inclined to care about an issue if the quality of their own experience is degraded by virtue of its’ existence.

This isn’t to imply some kind of failure on their part, or that they’re inherently “part of the problem” by taking this stance, it’s simply a fact of the human experience, given the sheer amount of problems and situations people find themselves in with only so much energy to dedicate to each. What that means in the context of GCAI, is something you’ve pointed out quite succinctly.

The real issue that can be cited with the use of GCAI for non-profit projects (Which is the vast majority of all content created and shared on this forum, and therefore is my primary focus for most opinions and discussions on the topic) is that it does create the unavoidable fact that with a lower skill ceiling comes a lower quality of output, both because of the quality of the tools used, or the lack of understanding or care on the part of the person creating it.

As a result of this increase in lower-quality projects, it unintentionally leads to the exact situation that creates the best breeding ground for hostilities to flare on both sides, but not for the reasons that would usually be cited. The simple fact is, low-quality, amateur projects made by a single person are easy pickings for people who want an example to point at in reference to the inherent flaws of the tools being used, and the moral failings of the people who use it, because of the negative impression it creates in the minds of people who view it. But just as much as it can be a scapegoat for people who view it negatively, it can also be used in a positive context to show the fact that even someone with little experience can still create something, even if poor by the standards of outside observers.

From my own experience with seeing this exact situation play out time and time again, this is exactly what I see repeated over and over. People are aware that there’s little recourse for their actions if they choose to attack someone with little influence or ability to retaliate, as opposed to an established or popular figure, and people who feel strongly to the contrary feel it as a moral imperative to defend someone they view as helpless against people they already view as fanatical or aggressive.

As I’ve stated before, my distaste tends to fall more towards a specific subset of anti-AI users, because of the clear acknowledgement through patterns of behavior and tactics that show that while they may believe in what they say, at the end of the day they recognize that their actions are more damaging than they are beneficial, and will resort to underhanded tactics when they feel they lack the upper hand to dominate others based off of their strength in numbers.

I don’t believe that people should be disallowed from stating their opinion, but it’s clear that the issue at hand doesn’t involve a discussion of opinions, it involves dealing with a certain group of people who have consistently proven unwilling or unable to communicate in a way that leads to any positive outcome. For the most part, these changes to moderation policy should lead to a more positive balance by providing the means to prune the people creating problems, if and when they do appear.

3 Likes
19

While I agree that most people using AI aren’t using it to try and make money, if you look at the recent projects posted a very large portion of the worst spammers/scammers are clearly just trying to make a quick buck.

A few bad actors can make a lot of shovelware, and right now there are no safeguards.

9 Likes
20

Even if I have ethical concerns about most LLMs in general, at this point I don’t really have an issue with people using it for projects and even charging for said projects. There are going to be people who use it well, and people who use it to churn out low-effort content, but the same is true of any user-friendly tools.

I do think that mandating an AI-generated-content tag or subtag (e.g., ai-art, ai-writing, etc.) on posts isn’t really the worst idea - it allows people to quickly filter past those things if they don’t want to see them. I know that there are a number of projects on this forum, even some high-profile ones, that don’t feature any disclosure about their use of AI-generated content, a lot of which isn’t obvious from screenshots, and for people with honest ideological opposition it’d be less than great to run into that after the fact. Whether it’d really cut down on the arguments about AI or just invite more of them probably depends on how heavy-handed the moderation changes around AI are.

7 Likes
21

I know some folks aren’t going to like who I’m going to be talking about, but some of these points are necessary:
1.) The real issues with AI is that they can be trained off someone else’s existing work and used to not pay the artist, there’s even a story someone had about a mother taking pictures of her work to put it through an AI, and didn’t even pay a single cent for the work she took photos of (we’ll come back to this later).
The biggest issue is that companies who most definitely could pay an actual artist for their work are actually using AI to, in a way, legally commit theft so they don’t have to pay the artist.
However, attempts like this are not something new: if a company thinks they can somehow take something from an artist without having to pay royalties or the like, they will find every possible way they can to circumvent it, whether by lawfare or otherwise.
2.) Something that I think most folks against it don’t understand is that things that’re being called AI are actually not, they’re tools that we’ve had forever that’re being called AI because that’s the hot new trend right now, either to gain venture capital or to conveniently blame for why certain folks at companies are getting gone. Which brings me to my next point…
3.) The thing people affected by AI should be afraid of isn’t the tool itself, it’s the people who use it: one issue for those outside of the art space (and even inside of it) is that the bigwigs at companies are being convinced that AI can do everything (the thought process for these types can basically be summed up at “I don’t know how this works, so it must be magic!”), so people are being kicked out without any real thought or understanding behind what the AI can actually do.
4.) One ethical method for all of this, if you have a style or ability but don’t have the time, is to get a local AI that doesn’t connect to the internet and train it off your own stuff: SomeOrdinaryGamers has a video on this. There’s also the benefit that if you have a system set up for this (apps that aren’t connected to a big server sending your data to the companies, non-spyware operating systems - like Linux, even setting up a secure VPN like using Mullvad - recommended due to the fact they don’t store your data, they just give you an account number that either need to memorize or write down - NOT ON YOUR COMPUTER OR PHONE),you can protect your own work and data as well from being stolen. However, if you don’t have the talents…
5.) There are several places that have art, music, programming, or what-have-you assets that are actually quite inexpensive, even free - it just requires looking around. I’ll make a separate topic on this later, but even game engines like RPG Maker, Unity, even Godot have reference sources and FULL LIBRARIES of assets you can use (where do you think AI is getting its training?). You can even use them to make your own stuff.
6.) There are cases where some folks in the space will wrongfully accuse others of using AI due simply to just better quality. This is even a concern at conventions. If you don’t want your concerns over AI to NOT be dismissed, don’t do this.
7.) This is probably something that A LOOOOT of people would want to hear on here, but there are ways to protect your work:

This ties back to that story from earlier, about the Art Fair lady: if you’re someone who sells their work in the real world, one method I would suggest would be:
1.) Have only your most basic works on display: no shadows, shading, not even colors, just use simple lineart for your display.
2.) Have your higher quality works either hidden or behind a curtain, with the rule that potential buyers can’t bring their phones, cameras, or any other possible recording equipment when viewing, since that means it can be put through AI.
3.) Have a rule-sheet on display and an explanation ready for why you don’t want your work recorded, and why you don’t want the buyer putting your work into an AI generator. If they try to argue against it, are clearly not listening to what you are saying, or they already decided to take photos of your work even with the rules on display, then politely show them the door or tell them to move along. You can even inform other creators at the convention of the behavior, maybe even the event organizers.

3 Likes