• sobchak@programming.dev
    link
    fedilink
    English
    arrow-up
    4
    ·
    1 day ago

    I think they could significantly lower their costs if they turn their focus away from the race to “AGI.” But, their valuations don’t really make sense unless investors believe they will achieve AGI.

    • Treczoks@lemmy.world
      link
      fedilink
      English
      arrow-up
      1
      ·
      16 hours ago

      I was there when the “Neural Networks” idea started. They thought they could program NNs to achieve AGI. Now they think they can do it with LLMs. But LLMs are just parrots with a large dictionary. They won’t reach that point, either. An LLM is way too much rooted in words to be able to think.

      • expr@programming.dev
        link
        fedilink
        English
        arrow-up
        9
        ·
        1 day ago

        How is this a response to a comment about AGI?

        LLMs are not currently, nor ever will be, anything remotely resembling AGI. AGI is still entirely within the realm of science fiction, like teleportation or time travel.

        • Jestzer@lemmy.world
          link
          fedilink
          English
          arrow-up
          1
          ·
          1 day ago

          Hey, we have AGI! We’ve had it for years! It’s called making babies 😎😎😎😎😎

      • Olap@lemmy.world
        link
        fedilink
        English
        arrow-up
        6
        ·
        1 day ago

        Yet still hallucinates mass bullshit and can’t count, can’t architect, and can’t write a decent test

      • sobchak@programming.dev
        link
        fedilink
        English
        arrow-up
        2
        ·
        1 day ago

        The “best” one I’ve tried is the latest Opus. I don’t trust any of them to use for real work, so I mostly just play around with a local Qwen 3.6 27B or Deepseek v4 Flash. I have heard OpenAI’s latest models produce less bloat than Anthropic’s.

        I should say I do find LLMs useful as a kind of search agent (both web and large unfamiliar code bases). And GhidraMCP is pretty cool (maybe just because I don’t have much experience with reverse engineering).