from the mod log:

Mojng5xTnVM7YHd.png

https://s.faf-pb.xyz/lXxek

The output from the LLM was provided as proof that someone needed to be banned.

I didn’t want to do this but my hand was forced by their dissembling, minimization and bullshit. Note how the output includes a nice big logo with a slogan and ascii graphics - this was not a one-off experiment, this is an app they were proud to call their own.

  • db0@lemmy.dbzer0.comEnglish
    1·
    2 months ago

    Another hit piece from rimu against /0. Must be a day that ends with -y…

    EDIT: Here’s rimu admitting that this is all based on a single modlog and him just making assumptions based on that and that alone.

    Here’s the series of events as I understand it

    • Rimu sees a two-week-old public modlog entry from unruffled banning a zionist as per our instance rules.
    • Rimu visits the public evidence linked and discovers that it’s a dump of the comments of said zionist, and it includes an LLM summary of those
    • Rimu sees the FAF logo on the pastebin. Rimu assumes that even though it’s clearly not a FAF domain, since it has the FAF logo, it’s an official FAF tool. This is false.
    • Rimu sees the LLM summmary and assumes that the banning happened due to the LLM summary. This is false
    • Rimu assumes that the tool that gathered these comments is automated. This is false
    • Rimu assumes that because it looks like an official tool, and despite the fact that he can only find a single modlog entry, all of the instance admins are regularly using it. This is false.
    • Rimu concludes that the instance is using an automated LLM judgement mechanism for instance moderation. This is egregiously false
    • Rimu does not reach out to any of our instance admins for clarification, even though we exist in the same instance-cooperation matrix channels. Even though our instance admins explicitly reached out to him earlier this month and asked to contact them for any questions/cooperation in order to avoid further drama and escalations. Note that there’s a reason why journalists ask for comments before publishing information.
    • Rimu doesn’t do any other due diligence.
    • Rimu rushes to publish with a cross-post to mastodon, because he knows the knee-jerk GenAI-hate there is much stronger than on threadiverse.
    • When called on any of the false and misleading statements he’s made by multiple people in the comments, Rimu completely ignores everyone countering him and all arguments, not to mention being so confused he doesn’t even know if he’s talking about mods or admins.

    I’ll leave the fact of why Rimu was randomly going through 2-week old modlogs of banned zionists as an interesting open-ended question.