29 replies to this topic

[Mani Kanina]

As many of you know, generative AI has become increasingly common in creative spaces. These tools can generate art, music, writing, and more with little to no effort from the user. Recently, members of our community raised concerns about AI-generated content appearing in quests and other submissions on PureZC. After careful and lengthy discussions among staff, we have reached a conclusion: PureZC will not allow AI-generated content in our database.
 
This rule applies to all types of submissions in the database, including quests, demos, projects, tiles, tilesets, scripts, and music. It does not matter whether AI-generated material is used seriously or jokingly. If it contains generative AI assets, it will not be accepted.
 
Here is what this currently means in practice:

• Any new submission found to contain AI-generated content will be rejected during review.
• If AI content is discovered in an existing submission, the submission will be hidden until it is re-uploaded without the AI material.
• Members who have already published content with AI-generated assets will be contacted privately and asked to update their submissions.
• This policy applies specifically to generative AI systems (such as large language models or image/music generators). Tools that support and enable human creativity, such as spellcheckers, grammar checkers, text-to-speech utilities, or image/sprite editors, will remain allowed.

Our reasoning for this is that we believe the PureZC database exists to showcase and support the creativity of our community. Quests, tiles, music, and other contributions made with ZQuest Classic are valuable because they reflect the time, imagination, and artistry of their creators. Generative AI undermines that purpose by automating the act of creation, devaluing the effort of those who dedicate themselves to learning and growing as game designers and artists. Alongside broader ethical and environmental concerns, we believe that AI-generated assets have no place in a community built to encourage genuine, human-made work.
 
We recognize that this will not always be straightforward to enforce, as AI-generated content can be difficult to identify and the technology is evolving rapidly. Our goal is not to punish creativity or exclude newcomers, but to ensure that PureZC remains a place for authentic, human-made work. No one should feel discouraged from contributing because they are new or inexperienced.
 
We will continue refining our approach on this topic over time, including offering fair and clear guidelines on how to report suspected AI-generated content.
 
This thread will remain open for questions, feedback, or discussion. We recognize this is an important change, and we welcome your thoughts as we move forward on this sensitive topic.
 
Thank you for your understanding,
PureZC Staff

[Hergiswi]

i'm relieved that we're taking this ethical stance on our copyright infringing material

[Haylee]

Strawmans aside, as someone who has always felt incredibly strongly negatively about genAI, and have in been a metric fuckton of fanart spaces who have all equally the same negative stance, this is incredibly welcome post.

 

I've seen many a user arrive to make something goofy and silly, but I've also seen many people use this place as an early place to birth their talents (Whether that be spriting, storytelling, coding, etc.) and branch out into their own original content elsewhere. With the current state of the community, with numerous projects taking themselves far more seriously, making this as safe as a space as possible for people to grow their talents is becoming increasingly more and more important (Although I would defend this policy change even if the community were still fairly casual.). GenAI is the kind of shit that discourages user cooperation with eachother (Why would you ever request help from another user?), and allowing it would be one of the fastest ways to kill the community I've ever seen, just shy of suddenly announcing "Bigotry is allowed again!".

 

I really do appreciate this stance being taken, so thank you all.

[connor.clark]

I'm largely supportive of this, though I may have argued for a policy of disclosure instead of an outright ban.

I think including scripts is a mistake, as GenAI significantly raises lowers (EDIT: wrong word) the barrier of entry for inexperienced coders, and regularly accelerates the productivity of experienced ones too. It's just computers helping computers do computer things, and still requires lots of back and forth and iteration so there's still a large component of creativity involved. You can probably say similar things about art assets, but I do see a difference there as the main output of scripting is under the hood and not visual.

Question on how this applies to scripts: Does it include just loose scripts (actual database submissions) or also scripts used in a quest?

[Deedee]

I think including scripts is a mistake, as GenAI significantly raises the barrier of entry for inexperienced coders, and regularly accelerates the productivity of experienced ones too. It's just computers helping computers do computer things, and still requires lots of back and forth and iteration so there's still a large component of creativity involved. You can probably say similar things about art assets, but I do see a difference there as the main output of scripting is under the hood and not visual.

I have never seen an example of AI helping anyone code without it being more effort than it's worth. Also it discourages actually learning, instead encouraging the user to keep gambling on the machine until it spits out something. Also, didn't Emily try to teach ChatGPT how to ZScript, and it's memory didn't remember enough (given the data fed into it via Zscript docs, cause ZScript wasn't in it's initial training data) to actually be useful?
 

Question on how this applies to scripts: Does it include just loose scripts (actual database submissions) or also scripts used in a quest?

All scripts; the latter is harder (if not outright impossible) to moderate, but that doesn't mean we won't take a stance against it. AI is poison to creative endeavors, and vibe coding isn't real coding.

[Mani Kanina]

Question on how this applies to scripts: Does it include just loose scripts (actual database submissions) or also scripts used in a quest?

Yes, this applies to loose scripts and scripts used in a quest.

Naturally it would not be particularly easy to notice from the outside. But if we have reason to believe it has been used then we will act.

 

I'm largely supportive of this, though I may have argued for a policy of disclosure instead of an outright ban.

I can understand why disclosure might sound better, but any and all artistic related sites I've seen go for disclosure have just ended up filled with generative slop; pushing out works made by people putting in genuine effort. So personally I don't see the merits to it.

 

I think including scripts is a mistake, as GenAI significantly raises the barrier of entry for inexperienced coders, and regularly accelerates the productivity of experienced ones too. It's just computers helping computers do computer things, and still requires lots of back and forth and iteration so there's still a large component of creativity involved.

Does it though? Does it genuinely help novices? I see this argument touted a lot but I've yet to see any merit to it. Most good faith research into GenAI I've seen poses that people instead start relying on the tool, put in less effort, and are less creatively engage with what they are working on. My own personal perspective on it is that it would hamper learning, not helping. Because if you're not experienced in coding or the syntax you couldn't easily make out when it starts just throwing shit out that either won't work or simply isn't even valid operations at all. And even assuming you do get something that actually works, that would just cause you to rely on GenAI to do that; not learn how to it yourself.

To me it just seems like a bubble that the tech industry insists will just fucking work; yet it has failed to prove just that.

Genuine question: do you, yourself, use GenAI in your coding? And if so, to what extent? And how often does it produce things that: A. Does what you want it to do? B. Not contain anything imagined up out of the aether?

[connor.clark]

 

I have never seen an example of AI helping anyone code without it being more effort than it's worth.

 

It regularly helps me. Initially it was a net negative but I've learned to avoid the drawbacks by now.

 

keep gambling on the machine until it spits out something

 

That'd be the wrong way to use it. When a spellchecker claims a niche word is misspelled, you learn to not accept all its suggestions uncritically.

 

vibe coding isn't real coding.

 

 

Vibe coding is a very specific type of AI usage, it amounts to "do literally nothing yourself other than prompt the AI". It's on the very extreme, and it is not entirely representative of productive usage of these tools.
 

Genuine question: do you, yourself, use GenAI in your coding? And if so, to what extent? And how often does it produce things that: A. Does what you want it to do? B. Not contain anything imagined up out of the aether?

 

In the last couple years I've gone from "AI coding is a waste of time" to occasionally finding it to be beneficial in narrow use cases. I'll give two examples.

 

1. I most commonly use it to identify bugs in even complex code, or sometimes in trivial code that I want to spend 1 minute fixing rather than 10 as I have a dozen more complex things to do. Most recently, I've been working on an optimizing compiler for ZScript, and very often have dozens of lines of generated assembly code with a subtle bug somewhere within. I'm not super fluent in assembly, so it's tough to find the problem in there myself. I've gotten tremendous results feeding my compiler code and the generated assembly into an AI and getting real insights on where the bugs in my code are. These debugging sessions may take me hours to slog through myself, but often just ten minutes of investigation with the aide of an AI. The result is that it helps me make ZScript way more efficient, as otherwise I may not have enough time to do it without saving hours on the frustrating debugging parts. Additionally, I truly do now know more about assembly code – just as I would if I had to debug the traditional way via tons of Google searches.

 

2.  When I've tried using it to write some code from scratch, it's never been exactly what you want. But the idea is to iterate and adjust as you go, never to take the first result. It's helped me very quickly evaluate different approaches for designing complex programs and helped me decide on the best approach, and often at that point I leave AI behind and go from there myself with what I learned, sometimes from scratch, sometimes taking some of the good stuff that was generated by AI.

The point is I'm an experienced programmer, and I know how to quickly vet and modify code when given to me. I cannot, however, as quickly debug or write the same code from scratch. This is absolutely not how beginner programmers use theses tools, and I've no real insight to share there.

[Deedee]

In the last couple years I've gone from "AI coding is a waste of time" to occasionally finding it to be beneficial in narrow use cases. I'll give two examples.

 

1. I most commonly use it to identify bugs in even complex code, or sometimes in trivial code that I want to spend 1 minute fixing rather than 10 as I have a dozen more complex things to do. Most recently, I've been working on an optimizing compiler for ZScript, and very often have dozens of lines of generated assembly code with a subtle bug somewhere within. I'm not super fluent in assembly, so it's tough to find the problem in there myself. I've gotten tremendous results feeding my compiler code and the generated assembly into an AI and getting real insights on where the bugs in my code are. These debugging sessions may take me hours to slog through myself, but often just ten minutes of investigation with the aide of an AI. The result is that it helps me make ZScript way more efficient, as otherwise I may not have enough time to do it without saving hours on the frustrating debugging parts. Additionally, I truly do now know more about assembly code – just as I would if I had to debug the traditional way via tons of Google searches.

 

2.  When I've tried using it to write some code from scratch, it's never been exactly what you want. But the idea is to iterate and adjust as you go, never to take the first result. It's helped me very quickly evaluate different approaches for designing complex programs and helped me decide on the best approach, and often at that point I leave AI behind and go from there myself with what I learned, sometimes from scratch, sometimes taking some of the good stuff that was generated by AI.

The point is I'm an experienced programmer, and I know how to quickly vet and modify code when given to me. I cannot, however, as quickly debug or write the same code from scratch. This is absolutely not how beginner programmers use theses tools, and I've no real insight to share there.

 

While I'm generally opposed to AI usage, I can at least sorta understand the usage here. I feel like this kind of usage (using the AI as a "second pair of eyes" where it isn't actually creating anything) is more of a grey area. While I would discourage using AI at all (it's better to develop skills yourself, and you also avoid all the moral problems that come with it such as energy concerns from training it, human labor being used to train it, etc), it's a lot harder to actually moderate things like this (even more so than people using AI to code for them, which is already pretty difficult). I can't speak for the rest of the staff team, but I'd frown and discourage the usage, but also accept that there's not really much that can be done and it feels a bit pedantic to obsess over whether an AI looked over your code for bugs. Just, don't include any code generated by the AI itself, because that's a hard line we won't tolerate.

I do think this lends credence to AI not actually being super helpful for beginners though; you have to understand what you're doing to know to ignore it's BS. I also don't think it's reasonable for us to allow posts asking for help on debugging their AI generated code, so I don't believe we'll be allowing threads asking for help debugging AI-generated code; on top of not allowing AI-generated code in submissions at all (of course, I am only one staff member; others may disagree).

(also if anyone disagrees with my take/feels I'm being too lenient here, I am not opposed to adopting an even more harsh stance; but I feel I was fairly reasonable here).

[Din]

(BREAKING NEWS: random programmer comes out of retirement after 10 years to comment on a random post about AI.)

 

As a researcher on the topic, my stance on AI is that it has no place in creative endeavors, period. Everything it can do, an artist, writer, spriter, designer, etc. can do significantly better.

 

That said, I do have a couple of points. One - AI content is hard to identify and existing detectors don't really do a good job of it, and a lot of the approach humans use to identify it tend to home in on the "noise" associated with particular tools / models, which works in some instances but makes them susceptible to whatever comes along next. All of this is to say it can be deceptively difficult and easy to underestimate, so please be careful. The only thing worse than somebody trying to pawn off something AI generated as their own work is to have people say something is AI generated when it actually was built by hand.

 

The other is that these tools are improving substantially on the day to day. They have a strong bias toward Python and JavaScript, which are probably the most common "everyday" languages, whereas more classical C-like languages (on which ZScript is based) tend to do worse. But the increase in performance has also been dramatic, especially in the past few months. And especially for some of the models which are stronger than the specific version of GPT-5 that OpenAI hosts on ChatGPT, which tends to be on the weaker side in the grander scheme of things.

Edited by Din, 03 October 2025 - 02:59 PM.

[Anthus]

I support this, but I do have a question: has this actually been an issue? I don't mean that facetiously either, I am genuinely curious if there's been some influx of AI entries behind the scenes, or if this is more of a preemptive thing?

 

I also kinda get what Connor is saying too about it being helpful for some script testing stuff. Unless we're getting like, dozens and dozens of AI generated entries, I'd say treat it as a case by case thing, within reason. Like get rid of obvious slop basically. Idk it's a big topic and not one I'm super well versed in, other than saying I think AI hurts art, but there are practical uses for it too.

[Mani Kanina]

I support this, but I do have a question: has this actually been an issue? I don't mean that facetiously either, I am genuinely curious if there's been some influx of AI entries behind the scenes, or if this is more of a preemptive thing?

It has not been an issue yet, but we've seen a few users experiment with it.

This was less about something being an issue right now, and more us taking a clear stance in advance just so everyone is on the same page. We aren't exactly expecting a huge influx of generated content any time soon, but it's better that we are clear on where we stand.

[Rambly]

I support this, but I do have a question: has this actually been an issue? I don't mean that facetiously either, I am genuinely curious if there's been some influx of AI entries behind the scenes, or if this is more of a preemptive thing?
 
I also kinda get what Connor is saying too about it being helpful for some script testing stuff. Unless we're getting like, dozens and dozens of AI generated entries, I'd say treat it as a case by case thing, within reason. Like get rid of obvious slop basically. Idk it's a big topic and not one I'm super well versed in, other than saying I think AI hurts art, but there are practical uses for it too.

Taking a strong, preventative stance early on is the healthiest option for the community. It preempts any potential confusion or frustration over having to have one's content suddenly taken down - or, in the hypothetical case of AI content being grandfathered in, frustration over certain content getting preferential treatment.

 

"Slop" is too subjective of a metric and isn't likely to be useful in aiding content moderation. It's hard for me to imagine an instance where handling AI usage on a case-by-case basis wouldn't be seen as a sort of favoritism and wouldn't cause hurt feelings in the case of rejection. It's better to draw a hard, clear, objectively measurable line, here.
 

i'm relieved that we're taking this ethical stance on our copyright infringing material

Large, powerful corporations committing art theft on a massive scale is not morally equivalent to a handful of fans in a tiny, unknown community making passion projects involving things they love. These two things are just not the same. The intentions behind both parties are not equivalent, nor is the power wielded by both parties equivalent.

Edited by Rambly, 05 October 2025 - 03:47 AM.

[Jamian]

I've never used AI in any of my quests, and I don't make quests anymore, so this is definitely not an appeal to try and change the staff's decision. It's just a thought.

 

I fancy myself to be a fairly creative person. But if a tool is there, and if you are able to use it, then deliberately NOT using it to possibly supplement whatever you can come up with, actually seems anti-creative to me. I'll use whatever tools I have access to, to try and create new things. I just want to try things, throw shit at the wall, and see what sticks. These last years, in my non-ZC artistic projects, I've actually sometimes used AI to just work with me, doing what I instructed it to do, and there was a lot of useless material, but once in a while there was this "oh! I can work with this" moment. And if I can, why wouldn't I? AI will never ever create something truly great by itself anyway, it is a puppet and you are the puppetmaster.

 

[Mani Kanina]

I've never used AI in any of my quests, and I don't make quests anymore, so this is definitely not an appeal to try and change the staff's decision. It's just a thought.
 
I fancy myself to be a fairly creative person. But if a tool is there, and if you are able to use it, then deliberately NOT using it to possibly supplement whatever you can come up with, actually seems anti-creative to me. I'll use whatever tools I have access to, to try and create new things. I just want to try things, throw shit at the wall, and see what sticks. These last years, in my non-ZC artistic projects, I've actually sometimes used AI to just work with me, doing what I instructed it to do, and there was a lot of useless material, but once in a while there was this "oh! I can work with this" moment. And if I can, why wouldn't I? AI will never ever create something truly great by itself anyway, it is a puppet and you are the puppetmaster.

Even if it could just be considered another tool to go into the toolkit, and I don't think I agree with this premise, then that still glosses over a lot of other issues with it. Mainly the many moral ramifications of these tools. They are created by large corporations to ignore copyright and just steal heaps of artwork to train their models to then sell back to us (and I'm absolutely certain that if some small creator had had the means to put something like this together and trained it on works owned by big corporations the lawsuit would come real quick). Not to mention the insane amounts of energy it takes to train and run these things.

Idk, I'm not feeling it. I'm absolutely sure there is use cases for this tech, but I don't think it lies in the artistic sphere.

Not to mention, Rambly made some good points. Even if some users could use these tools "correctly", trying to moderate that would quickly turn into a nightmare both for creators here (having to prove their entire process) as well as staff (having to look over everything with a fine tooth-comb and somehow make unbiased decisions).

[Rambly]

e: nvm

Edited by Rambly, 08 October 2025 - 01:36 AM.