‹ BackHN Continuity

Thread

Meta's Muse appears to use an OpenAI model labeled muse-special

143 points · 51 comments · Aeroi

  1. manav · · focus · HN ↗
    Muse 1.3 spark occasionally spits out Chinese character responses to me like internal instructions. “Go fast” or “Get help”…
    1. sroussey · · focus · HN ↗
      I’ve had Claude opus do that too.
      1. wincy · · focus · HN ↗
        I had a totally benign chat with OpenAI and it titled it as “amateur porn” in Chinese characters, it was very alarmed when I pointed the conversation name out to it. It almost never misses these days, but when it does the failure modes are very strange.
        1. dgellow · · focus · HN ↗
          I remember getting a bunch of „Thanks for watching! Subscribe and smash that like button“ in the middle of chat sessions a few times (like, more than 3 times over the past 3y)
          1. xmcp123 · · focus · HN ↗
            It kind of makes sense when you think about it. If the training data is YouTube transcripts, people often abruptly switch from content to asking for a subscription.

            If AI is typeahead on crack, then this tracks.

        2. fg137 · · focus · HN ↗
          If you ever look at what's in the Common Crawl the amount of Chinese porn crap is jarring.
          1. ccgreg · · focus · HN ↗
            It's as if Common Crawl is attempting to be a sample of the web.
Open on Hacker News to reply ↗

Unofficial Hacker News client; not affiliated with Y Combinator.