• 0 Posts
  • 54 Comments
Joined 2 years ago
cake
Cake day: June 4th, 2023

help-circle









  • OhNoMoreLemmy@lemmy.mltoFediverse@lemmy.worldHappy #GlobalSwitchDay
    link
    fedilink
    English
    arrow-up
    4
    arrow-down
    6
    ·
    5 months ago

    Wasting other people’s time.

    If you want to use an LLM that’s fine, but if you’re cutting and pasting it into a discussion you should warn other people that it’s not human generated.

    And most of it isn’t wrong, it’s just a giant wall of text that’s largely irrelevant to the conversation.






  • I mean the fundamental problem is that humans are dicks and moderation is always needed. It should also be paid, and supported with counciling and recovery time when needed. Dealing with toxic content is a job.

    Federation isn’t very good at this. The tech is great but everyone is a volunteer and there’s (afaik) no global ban hammer so trolls move from one instance to another. Bluesky currently has venture capital to pay for moderation teams, and centralized ban options.

    I don’t know how long this can last without advertising revenue though.




  • This probably isn’t a hallucination in the classic sense.

    This is probably a near copy of a forum post where a user was channeling fight club and trying to be funny. The same as the putting glue on pizza thing.

    And guardrails don’t work very well. They’re good at detection tone but much worse at detection content. So an appropriately guardrailed LLM will never call someone a “fucking ######” but it’ll keep telling everyone that segalis have an IQ of 40 until there’s such a PR backlash that an updated is needed.