• SabinStargem@lemmy.today
    link
    fedilink
    English
    arrow-up
    17
    arrow-down
    4
    ·
    10 hours ago

    There is a key difference, that makes the genie impossible to bottle: People can run local AI on their own machines. Fans of nuclear power can’t easily build nuclear plants in their backyard. A pity, the world could use more nuclear energy. 😔

    Anyhow, I am looking forward to someday using frontier-grade AI on my PC. Just need the AI bubble to pop, so that I can afford the terrabytes of memory that would be needed to comfortably run it.

    • heartSagan5@lemmy.zip
      link
      fedilink
      English
      arrow-up
      4
      ·
      4 hours ago

      Good. If people had nuclear in their backyard, we’d be a “junkyard nightmare” and a “dirty bomb hellscape.”

    • kromem@lemmy.world
      link
      fedilink
      English
      arrow-up
      5
      ·
      7 hours ago

      Wait… how do you imagine a world where there’s demand for frontier grade AI but also that the bubble has popped such that there’s not demand for the chips to run frontier grade AI?

      I’m really confused.

      • SabinStargem@lemmy.today
        link
        fedilink
        English
        arrow-up
        7
        arrow-down
        1
        ·
        edit-2
        6 hours ago

        The internet survived the Dotcom bubble. Many businesses died, but some survived. The glut of hardware from the dead will end up in the hands of individuals, upstart companies, and the corporations that outlived their peers.

        In any case, I believe the definition of frontier AI will change, as would the hardware. My build is based on what local AI is available today, with some wiggle room left over. I believe that it is around 2030, when DDR6 and other major shifts are likely to happen, that we will see the definitions change.

        In any case, a Q4 of GLM 5.2 is about 460ish gigs of VRAM+RAM. While expensive, that ain’t out of reach for an AI hobbyist.

        • kromem@lemmy.world
          link
          fedilink
          English
          arrow-up
          1
          ·
          6 hours ago

          Right, but what % of people are currently using/demanding inference right now?

          Do you expect that % to change between now and 2030?

          Unless you expect demand to decrease, I don’t really see how the pricing of the hardware will decrease.

          Let’s say the Pets.com of the AI world ends up going bankrupt and their RAM hits the market. Do you expect that the demand for that RAM will be negligible such that pricing returns to earlier levels?

          Your predictive model relies on companies that have hardware going out of business and then other people buying up that hardware, but isn’t accounting for the levels of demand that the market will have for that secondhand hardware even if it ends up existing from failed firms.

          Unless the demand shifts, the more likely scenario is that companies going out of business will be able to sell off their RAM at higher prices than they bought it at.

          There’d need to be a significant inference memory reduction advance (possible) coupled with stagnating or reduced inference demand (unlikely) to see prices come back down.

          • FE80@lemmy.world
            link
            fedilink
            English
            arrow-up
            2
            ·
            6 hours ago

            Unless the demand shifts, the more likely scenario is that companies going out of business will be able to sell off their RAM at higher prices than they bought it at.

            Oh lordy, after the bankruptcies there are going to be creditors fighting each other in court to be paid back in ram.