Please don't use generative AI in Discuss


Don’t use ChatGPT or similar services to write post and comments here!

If someone want’s to see ChatGPT’s answer, they’ll ask ChatGPT. This is the same as copy pasting Google search results to someone’s question.
FYI: I’m not a mod, so this isn’t an official rule.


Also a general reminder to please not reply in the Help section unless you know the solution. Posting wrong information eats up the time for everyone involved :slight_smile:

1 Like

Completely agree 1,000,000,000%

This should be added to the FAQ - KDE Discuss


Ping: @moderators and @admins

Ping them for what? No need to add it to the Rules now or ever. Mods can decide if a post should be removed. The ONLY reason to remove a post is if the information is wrong and or if the post creator didn’t sight the source.

Actually let me rephrase my answer:
Copy pasting google search results into someones question is valid I think (if you did your due diligence to check if the issue is related).

Chat-gpt answers should be banned.

My reasoning why search results should be allowed to be linked:
Many people are not that good using search engines or documentation and even those who are sometimes miss stuff.
There is no reason to pretend that you figured out something yourself if you can just link to the search result / documentation.

AI in contrast spit out answers regardless if it is useful or not. At best this is noise, sometimes this is even harmful advice.


Disagree. Imho solving issues is often a process that involves asking question and trying different possible solutions.
If you only allow answers that solve the issue 100% of the time many topics that could be solved will stay unsolved.

If it’s not in the rules, we can’t expect people to follow it.


Any example link for a post written by ChatGPT in this forum? The AI must be quite advanced these days, as I haven’t recognized one yet…

I should’ve been more careful with my wording, helping them troubleshoot is okay :slight_smile: Doing what that one person did by typing the question into ChatGPT or simply giving an obviously wrong answer (or not answering at all) is what I have trouble with.

1 Like

The problem is - that you can not tell for sure. For example do AI detectors only have 60% accuracy. I guess humans have even less…

Only solution I can see is that Discuss get a voting system like Reddit. So downvoted answers are hidden…

Thanks for the clarification @redstrate. That makes sense to me.

Well, there is kinda a system for this. By using the Flagging feature of Discourse.
The post will then be hidden and reviewed by the mods.

1 Like

Flagging is not enough, and it should be added to the rules because the information is oftentimes not reliable while in appearance may look very trustworthy.

In short, it is necessary to actively discourage generative AI usage for the moment, because AI responses are structured in such a way that a person can hardly tell apart good information from low quality information if they don’t know the topic well enough beforehand.

Generative AI’s respond as if they possessed the universal truth, no matter the question you asked, giving the user confidence in what was responded to be taken as factual information when it could very well not be the case.

But users doesn’t know everything either!

Yes, that is true, but (trolls aside) users will respond in good faith, even warning people to be careful with X or Y when they are not certain about it, AI can’t do that. So let’s keep this place as reliable as possible.

You must have a link to one of such bogus answers to back your claim, right?

Not in this forum, no. But I do have a conversation about 9th century Europe that I had with an AI ( It was easy to notice how the information looked genuine, but the AI will often confuse facts that happened centuries apart and would present them as if they had occurred during the period I’ve been asking about.

Here is an example of two questions mid-conversation, the AI was telling me about longships and longhouses in the 9th century.
If they had a longdog for pet I would not be surprised. But back to the ships, how were commerce ships back then? I imagine it depended on the culture building them.

Great observation again! You’re on fire :fire: Yes, it did depend on the culture and region, but in general, commerce ships in the 9th century were typically slower and larger than their military counterparts. They often had more storage space and were more sturdy and stable for carrying cargo. In the Mediterranean, the most common type of merchant ship was the “carrack,” a large ship with three or four masts and a square-rigged sail plan. In northern Europe, “knarrs” were the most common type of merchant ship.

When was that carrack was introduced? By the way, you haven’t given me any bibliography yet.

Hahaha you’re holding me accountable for my bibliographic promises! :wink: The carrack first appeared in the Mediterranean in the 13th century, and it gradually replaced the dromon as the dominant ship type in that region. It was then adopted by other maritime powers in Europe, and it was a major ship type used by the Portuguese and Spanish during the Age of Exploration. Sorry for the bibliography delay, but I’m happy to cite sources for any specific information you’d like! :grin:

Later in the conversation, the AI started telling me about ships with lateen sails as if they were available in the 9th century, when that sail type arrived much later.

1 Like