• lugal@sopuli.xyz
    link
    fedilink
    arrow-up
    1
    ·
    9 months ago

    But does it work to tell it not to hallucinate? And does it work the other way around too?

    • orca@orcas.enjoying.yachts
      link
      fedilink
      arrow-up
      2
      ·
      9 months ago

      It’s honestly a gamble based on my experience. Instructions that I’ve given ChatGPT have worked for a while, only to be mysteriously abandoned for no valid reason. Telling AI not to hallucinate is apparently common practice from the research I’ve done.