Don't Use AI for Replies

Eric Chadwick · Jan 7 2026

Hi there! This came up recently in a couple places here on the forums, so it makes sense for us to make a clear policy on the use of AI search engines like ChatGPT, Gemini, and the like.

Let us know what you think about this. Your thoughts matter, since in the end the contents of these forums are purely built by all of you, each and every day. You are Polycount!

https://polycount.com/discussion/63361/information-about-polycount-new-member-introductions/p1#don-t-use-ai-for-replies

Don't Use AI for Replies

Polycount doesn't allow the use of AI tools to help you generate answers.

We’ve seen it again and again, these algorithms snip and combine sources, often producing inaccurate results, because they lack the necessary context. Then they present their results as facts, and most users accept them without seeking (clearly necessary!) verification.

If you must use these tools, use them smartly. Don't accept their output at face value. Verify sources if provided. If no sources are linked, then it's best to simply treat the answer as pure bullshit. Seek to verify the output by reproducing it manually yourself.

If you see AI-generated text in replies, please use the Flag: Report function to let us know. We want the forums to be a source of usable information, not a bunch of AI slop.

Rima · Jan 7 2026

I like this policy. One of the biggest things Polycount has going for it, to me, is the fact that you can come here and talk to people who know their shit. They can talk about industry stuff as someone in it who's experiencing it, talk about technical stuff because they actually work with it, give feedback on artistic stuff because they're artists, etc. AI, on the other hand, doesn't know shit and will confidently bullshit you with useless crap that won't lead you to become better at anything. It's basically worthless. And that aside, it's just too sad to replace actual communication between people with soulless robots.

This is definitely the right move.

MikeF · Jan 7 2026

Thank you, this is great

Eric Chadwick · Jan 8 2026

I suppose we need something about AI-generated assets.... concepts, textures, meshes, rigs, etc. But I'm not sure what the line is here.

Some form of AI/LLM is in many tools now, sometimes just as a search engine, but often as a next-step recommendation tool.

sacboi · Jan 8 2026

Yeah great stuff and in relation to ai supported tool development or partial integration thereof, perhaps an advisory of some sort for vendors / solo devs, recommending full disclosure?

In my opinion, I think it's only fair because people do occasionally promote their product/s on this forum so rather than omission, simply introduces clarity for a potential end user to either then opt out, trial or indeed purchase.

zetheros · Jan 8 2026

yea, disclosure that AI was used sounds good. As for the rest, I can really only speak to what I see and do on a daily basis. After a day's work of making 3d assets for my job, I spend an additional hour or more making art on the side, and this is what I've done consistently for the past 3 years. During that time I have yet to find a genuinely compelling reason to use generative AI.

gnoop · Jan 8 2026

Isn't this forum exists to express your thoughts freely without polishing and censuring yourself. I am pretty tired of reading that overly polite, inexpressive and graphically blind language Ai is typically promoting . It's sleepy and dull after all.

Well , until you specifically ask otherwise but nobody do usually .

Whats the point of using AI here for reply? My guess when your english skill is bad and it's only way. Not a crime in such case probably.

Eric Chadwick · Jan 8 2026

gnoop said:

Isn't this forum exists to express your thoughts freely without polishing and censuring yourself.

Things are generally pretty loose around here, but also it's not just post whatever nonsense you want, we do have some rules, and all of them exist for good reasons. If you have some really good output from an AI tool, I'd love to see it. Most of what I've seen is not great.

It seems to me when people seek info in an area they are not a subject-matter-expert of, the AI result is crap, but the person doesn't have the understanding to discern that it is indeed crap.

gnoop said:

I am pretty tired of reading that overly polite, inexpressive and graphically blind language Ai is typically promoting . It's sleepy and dull after all.
Well , until you specifically ask otherwise but nobody do usually .

Sorry, I didn't understand this, what is this referring to exactly?

gnoop · Jan 9 2026

Eric Chadwick said:

Sorry, I didn't understand this, what is this referring to exactly?

Well, I meant that specific corporate or showcase language I am feeling sleepy reading and listening . Super polished. Overly positive, corners rounding. Seen it in many public performances , developers announces on youtube etc I couldn't see through till the end . I blame AI for this but maybe I am wrong. It just coincides with a rise of AI . Before I haven't noticed it much. A certain speech style. The end of anything live and spontaneous which often makes things interesting to see, listen or read . Even when a person shows some speech awkwardness or odd humor. Not that I want a Shakespeare's drama and sharp language here but once everyone starts to polish texts through AI i feel a tad bit of boredom. Seems half of youtube is AI already.

Just an impression nothing more . And I am puzzled why would anyone even bother to use AI for forum reply and hide the fact . I don't want back the forum flames here but to be perfectly honest they were entertaining to read and comment.

I agree that discussing what info and content AI create Is helpful , good or bad . But usually it's totally another case , not exactly using AI for replies.

Celosia · Jan 9 2026

Corpo-speak has been around for quite a while, LLMs only made it more accessible. So we see it everywhere now, accompanied by weird emoji listicles when it's a professional seeking work. It's text shaped in a way that should be more legible in principle but isn't in spirit because it's created by a hollow bot charging by token.

But I can see a use case for LLM rewrites: Non-native English speakers.

Being able to read and write in English doesn't automatically translate to being concise and accessible. I say that as a non-native speaker. Depending on your mother tongue your sentences can get very contrived, and LLMs could be used as a tool to extract what you meant to say from what otherwise are very winding slabs of text.

That's a usage much closer to a spellcheck than what's being discussed in the rule because the bot isn't creating the contents of the text, and personally I'm not keen on it due the lack of spontaneity thing, but I still find it acceptable because then it's actually using it as a tool with the end goal of conveying your thoughts, not as a way to offload thinking to a deceptive impression of a brain.

Regarding AI-generated assets I'm on the full disclosure team, and also "it should not be the final asset" team. Using tools that happen to have ML is a gray area, but doing something like generating slop to pass it off as concept art for example, the end goal for a concept artist, is very different.

poopipe · Jan 9 2026

I believe the spirit of this is ... post your own opinion/answer.

using a robot to translate your words isn't wrong,

Zack Maxwell · Jan 9 2026

zetheros said:

yea, disclosure that AI was used sounds good. As for the rest, I can really only speak to what I see and do on a daily basis. After a day's work of making 3d assets for my job, I spend an additional hour or more making art on the side, and this is what I've done consistently for the past 3 years. During that time I have yet to find a genuinely compelling reason to use generative AI.

It blows my mind that no tool exists for AI-automated retopology.
People have completely jumped ahead into making shitty full AI-generated models which are ugly and completely unusable for anything, but nothing for retopology