This is a reminder to all forum users of the existing artificial intelligence and chatbot policy. AI-generated compositions, whether text or image, are subject to summary deletion by any member of the @mods-general per that policy. We have been advised to use Pangram to test texts for probability of having been generated by artificial intelligence and will apply that standard.
Continued violation of the policy may result in additional sanctions, including suspension from the forum, by vote of the mods-general as a group, per existing operating procedures.
As far as I’m aware from other people who used that and other services, because exactly it gives a possibility that the given text is AI-modified, doesn’t mean that even the 100% is true. It gives a strong indication for sure, but it’s still isn’t certain.
Because of this uncertainty, shouldn’t the usage of that service be as an additional guide to help determine whether a message was touched by AI, rather than being an absolute indicator?
So, does this mean that, from now on, non-English-speaking contributors will be permanently prevented from using LLM-based translation to communicate with English-speaking contributors?
In principle, I support restricting documents that are generated by LLMs. However, I had hoped there would be a clear distinction between content authored by an LLM and the legitimate use of LLMs as a tool.
In my case, I have often been mistaken for having written my posts with an LLM simply because I write them myself in my native language and then use an LLM to translate them, in an effort to preserve the subtle nuances of what I intended to say.
Does this policy mean that even non-English-speaking contributors who use LLMs merely to communicate in English—or to compensate for their limited English proficiency—will also be blocked?
Or will contributors now be expected to prove that every English post they write was actually written by themselves?
It should be imo because some so-called AI detectors also use AI themselves (and whatever it was trained on), so text that’s been public domain since god knows when might be flagged as AI
Please try to understand just how awkward and ineffective the forum’s built-in translation can be for people who don’t speak English. It often fails to convey the intended meaning.
Even translations generated by LLMs can sometimes be inaccurate. Do you really believe the forum’s built-in translation is sufficient?
Both are automated translations. You can argue the forum should use a different translation provider, such as a LLM.
Just imagine you are writing with a French guy. Why use English which both of you don’t understand and feed the text in another LLM. Doesn’t make any sense.
If the linked Policy on Chatbots and Artificial Intelligence is indeed the only policy regarding Artificial Intelligence, then using an LLM to translate your original text into another language to aid meaningful and frictionless discussion would not appear to be in violation of OSM policy. If the tool that the moderation team uses is any good, I doubt that such passages would flag as AI.
Hopefully the team will be courteous enough to clarify their position, as I agree that you would have a perfectly acceptable use case to enhance yours and others’ experience on these forums.
At the risk of getting myself sanctioned, does this apply to moderators too? You posted this clearly AI generated text from Copilot yesterday:
If you want people to follow the rules and not post AI slop, then you shouldn’t be doing it either. As a moderator, you should be setting an example for everybody else.
But prompting a chatbot to write a message for you also often fails to convey the intended meaning.
Plus the issue seems to more often be in opposite direction: people who don’t speak English well trying to generate English text with an LLM. Then they night not able to tell if it says what they wanted to say…
I can’t speak for all languages, but anecdotally, the forun built-in translation is quite good for Polish, which is not one of the world’s major languages. German too, though it has more speakers.
I’ve just read through that thread. While I don’t at all understand the technical outline being discussed, it’s disheartening to see the community dogpile on and so brazenly condescend a user attempting to clearly explain and expound upon their position—especially when the accusation of AI use may appear to be misguided.
I have no interest in sparking again any argument, but I thought it best to make my opinion known as a new member of this community (and what little that may matter).
I’m fairly sure they’re referring to the thread that was linked above, which sparked this whole AI discussion. Multiple forum regulars being [expletive deleted], just a normal day here really.
Now
This isn’t the first time adreamy has explained their position, and some of you just keep telling them they should do what you want instead.
Adreamy’s position is about agency. They know most people here would completely rely on automated translation to read whatever they wrote in their own language. If they are the one that provides that translation, they can do QA to see if it says what they want to say.
They would have no control over what the forum translation or some other tool does. At worst they’d have different people arguing back based on wildly different translations, and they wouldn’t even know what those translations said.
You can disagree if you want, but it’s a valid position to take.
You’re entitled to that opinion, but I’m fairly certain that, in the thread about Wikimedia Commons MIDs, I spent more time reading and thinking about how to reach an understanding with the original poster than they spent responding. I have no proof, other than the seconds it took for them to throw another wall of condescension my way each time I said something. Maybe others had different experiences.
For those unfamiliar with Discourse’s moderation tools, what this looks like in practice is that the post gets a icon at the top, indicating that it has been edited (just like when someone edits their own post after some time). You can click the pencil to see what the moderator has deleted from the post. If you’re coming to a moderated post after the fact, you may need to dig through these diffs to understand why others reacted how they did. Otherwise, the reactions would seem quite over the top!
Wasn’t referring to you, for what it’s worth. But I guess you know exactly which post of yours I might have meant, and maybe you’re not exactly proud of it. Because you don’t come across as someone who’d enjoy posting like that.
icon at the top, indicating that it has been edited (just like when someone edits their own post after some time). You can click the pencil to see what the moderator has deleted from the post. If you’re coming to a moderated post after the fact, you may need to dig through these diffs to understand why others reacted how they did. Otherwise, the reactions would seem quite over the top!