Reminds me of this:

I think atproto is a good protocol, but god bluesky-the-company is dogshit.

  • Trainguyrom@reddthat.com
    link
    fedilink
    English
    arrow-up
    4
    ·
    2 days ago

    Yeah that was my biggest takeaway is these posts seem to assume sentience in what’s little more than a sophisticated “most likely next word” generator. There’s tons of cool things that can be done with these new machine learning tools, but they are not sentient, they are not close to sentience and we may never invent artificial sentience.

    The one thing we now know for sure is we can damn well convince people of sentience artificially far more easily than I ever suspected

    • shalafi@lemmy.world
      link
      fedilink
      English
      arrow-up
      2
      ·
      18 hours ago

      We’ll get true AI one day, but the timeline and methods are up in the air. I’m not even sure LLMs will be a piece of the puzzle. Guess we’ll learn something from the exercise.