Although I don’t have time rn to engage on the topic, I thought it still worth pointing out that there doesn’t appear to be much of a control mechanism for bots joining the forum and contributing comments.
FullVINQubes Posted 51 mins ago / Joined 17 hours ago / Read 27m
27m of Read time per 19 seemingly relevant posts is quite the superhuman ratio…
also “Trust Level basic user”.. Perhaps Trust Level should include a bot category?
I second that. I’ve also been puzzled by some posts where someone first agreed, only to then ignore any objections and write more nonsense.
Maybe human heavy used LLM to generate his answers.
Are bots allowed? Is it something we want? Discourse tries to prevent bots and it works, most of the time.
Thanks for the feedback. I’m sorry for this mess: it wasn’t flagged as a bot and I had a hesitation while reviewing their first posts, just before it became clear it was a bot. Their goal was to promote a VPN provider. I guess the generic LLM-generated posts are a way to look like a regular user? Now I don’t think I have a way to do something without clicking on every post. Some users already flagged some of the posts, if you see something really suspicious, please flag it.
Side note: I don’t think it is a good idea to look at the Discourse stats about “read time”, etc. One could browse the forum without using their account, or with the mailing-list mode, write their answers in an offline file and just log in only to copy-paste that.
Think that content from LLM should be allowed, with disclaimer “this information provided/generated by LLM”.
Example: Disposible Qubes DoS Incident - #3 by WhiteShadow "Answer by AI module." and “End of AI content”.
Contributions to the Forum are subject to our general guidance on use
of AI in contributing to Qubes, as set out here
Specifically:
If you use generative AI in your contributions, you must disclose this
at the time you submit your contribution.
Regardless of how your content was created, you will be considered the
fully-accountable author of that content, and you are responsible for
reviewing & fixing such output before submitting it.
Likewise. I don’t want bots that push content onto the forum, even high quality content, because it would amplify some voices over others. However, I wouldn’t mind a built-in search assist bot that could summarize multiple relevant posts, with attribution, but without creating a new post.
I think the Qubes AI policy covers the AI assist case well enough. I am concerned that it doesn’t sufficiently address the case of an AI agent independently joining the forum and pushing content, without a clear chain of responsibility.
There is no chain of responsibility for an AI agent, and an AI agent is
not able to share an interest in contributing to Qubes, or anything else.
Since they are unable to do this, they have no place in the Forum.