Weve been going back and forth on this in the moderator team for a while now and its about time we said something publicly rather than leaving everyone guessing. AI isnt going away, a fair few people here already use it in one form or another, myself included, so pretending otherwise would be daft.
The short version is that using AI isnt the problem. Posting things you cannot vouch for is. Everything below comes back to that.
No AI agent accounts.
Accounts being run by a bot rather than a person are not welcome here and will be banned. We have had auto reply bots on here before and its just pollution. People come to a forum for help from other people. If anyone wants to talk to an AI there are plenty of sites for that, it doesnt need to happen here.
Dont paste AI output as an answer.
Anyone can log into ChatGPT and ask a question, theres nothing clever about it and it doesnt help the person who asked. AI will happily tell you that a chip does something, or that a fault is caused by X, and it reads convincingly, but most of it is suggestions and theories that nobody has ever actually tested. Its usually very long and nobody else in the thread can check any of it either.
The reason we care is that it doesnt stop there. Somebody reads it, takes it as fact, repeats it somewhere else, and eventually it gets scraped back into training data and comes back round as "fact" later on. Weve all seen wrong information about the ST repeated for thirty years already without any help. This speeds that up enormously.
If you use AI to look something up, thats fine, thats what its there for. Read it, check it, then tell us in your own words. This is really just the Discussions rule further up the page, the forum isnt the place for guesswork, speculation or hearsay. AI is simply the newest source of it.
If you cannot vouch for it, dont post it as fact.
This is the actual test, rather than whether an AI was involved. If you post something as fact you should be able to stand behind it, same as we would expect of anybody quoting anything else.
That doesnt mean prove everything before you open your mouth, none of us could manage that. It means dont make something sound more solid than it is. If you tested it, say so. If you read it somewhere, say where. If AI suggested it and you havent checked it, say that too, and keep it to a line or two as a lead for somebody else to follow up. Thats genuinely useful. Ten paragraphs of untested AI theory presented as an answer is not.
AI as a tool is fine and doesnt need declaring.
Spell checking, grammar, translation, tidying up your own writing, code debugging. The knowledge is still yours, the AI just helped you get it out. Nobody needs to declare any of that.
Part of the reason is practical. If everybody who used AI to sort out their spelling had to stick a disclaimer on the top, half the forum would be carrying one, people would start skipping those posts, and the only people it would actually affect are the honest ones who bothered to say. It achieves nothing.
The other reason is that its not really as new as people make out. The argument goes that an LLM writes the code for you and other tools dont, but that was never true. BASIC is wrappers round assembly. C coders have never declared which prebuilt libraries they pulled in, and thats somebody elses work doing the heavy lifting just the same. Assembly, C, BASIC, LLM, its all the same ladder and weve just gone up another rung.
Remember Prehistorik? Everybody said it was brilliant, right up until word got round it was written in STOS, and then people turned their noses up at it. Same game, same code, nothing about it had actually changed. The only thing that changed was people finding out which tool made it. Its the same with soldering. Everything used to be done by hand, now its machine assembled, and nobody complains about that when they see what it did to the prices.
Worth being clear that this only stretches so far though. It works for code because code proves itself. It compiles or it doesnt, it runs on real hardware or it doesnt, and anyone who doubts it can go and test it themselves. A wall of AI text about how some chip works doesnt run and nobody can test a word of it. So this isnt a defence of slop and please dont try to use it as one.
We judge posts, not accounts.
Nobody here is getting moderated over how they write. Poor English, translated text, odd phrasing, long posts, tidy grammar, em dashes, none of that is evidence of anything.
Theres no reliable way to tell AI text from somebody using a translator, or somebody who just writes a bit oddly, or somebody who finds writing hard going. If we start guessing we will get it wrong, and we will end up accusing real members and losing them. That would do far more damage to this place than a bit of AI slop ever will.
For the record, what we do look at is behaviour rather than style. An account that only ever supplies answers and never once has a problem of its own that it needs help with. Strong opinions stated as fact from a brand new account with nothing personal in any of it. No interest in anybody elses reply, no shifting position, criticism never addressed. Even then thats a nudge to look closer, not proof of anything, and its our job rather than yours.
Report, dont accuse.
If a post concerns you, use the Report Post option and a moderator will look into it. Please dont call it out in the thread. This is the same as the existing reporting rule further up and its there for the same reason. Once threads start turning into arguments about who is or isnt a bot, the thread is finished and so is any chance of the original question getting answered.
Declaring AI generated work, a guideline rather than a rule
If you release a project which used AI generated artwork, music or similar, please say so. "I used AI to generate this artwork" rather than "I made this artwork". "AI and I wrote this program" rather than "I wrote this program".
Its basic common courtesy, and without it the forum starts looking like its full of expert artists and coders when it isnt, which isnt fair on the people who genuinely can do it.
Its not a rule because we cannot enforce it, and theres a worse problem with making it one. Anybody who doesnt care simply wouldnt declare anything, so the only people it would ever land on are the honest ones who did. Thats the wrong way round. So it stays a guideline and comes down to your own conscience about claiming work you didnt physically do.
What we are not going to do
- Ask everybody to declare all AI use, for the reasons above.
- Declare the forum an AI free zone. Id be breaking it myself the same day with the server write ups and code debugging, and Im not having a rule Im exempt from.
- Add more hoops to signup. Tickboxes, questionnaires, puzzles. None of it stops anybody who was going to lie anyway, and the cost always lands on real people trying to join. Weve all seen forums get so locked down that the genuine members cant get in either.
- Run posts through an LLM to score how likely they are to be AI. It will tell you a post is 95% AI with total confidence when it isnt, and we would have no way to check. Using the thing everyone is complaining about to make decisions about real people isnt a good look.
- Treat punctuation or writing style as evidence of anything.
Being honest about the bits we havent got an answer to, because this isnt finished and Id rather say so than pretend its watertight.
- Enforcement. We can write "can you vouch for it" down but we cannot actually prove anybody didnt verify something. Realistically its something we lean on when a post is obviously a paste job, not something we can police properly.
- Where a project counts as leaning heavily on AI. Artwork and music are easy enough, code is murkier. If somebody AI generates an entire game, does that need declaring? Code proves itself by running so we lean towards no, but its not clear cut.
- Somebody using AI for their entire side of a conversation is misrepresentation, but theres no way of spotting it that doesnt turn into judging people on how they write, which is exactly what we said we wouldnt do. No answer to that one yet.
- Consistency between moderators. If three of us have different ideas about what counts as slop then youll get inconsistent moderation and rightly complain about it. Something we need to work out between us.
- People who object to AI on principle. For some its the ethical, social and environmental side rather than anything to do with accuracy. For others its about craft and process rather than just output, and no interest in something knocked out by somebody who doesnt know or care how the hardware works. Neither of those is a complaint about content, so no content rule fixes either of them. Whatever we settle on, some people wont be happy with it, and theres no version of this that keeps everybody. What we can do is not rub anybodys nose in it and make sure this place doesnt turn into wall to wall AI chat, so those members still have a reason to be here.
- The bigger picture. None of this stops forums generally filling up with AI conversation over the years. Rules help at the edges. The wider problem isnt something we can fix at forum level.
Nothing here is set in stone either. If you think something is wrong or missing, say so in the thread.