• Blue_Morpho@lemmy.world
    link
    fedilink
    arrow-up
    1
    ·
    edit-2
    2 days ago

    It’s different because a camera didn’t need to generate enough carbon to burn down an acre of rainforest to take a single picture

    Absolutely false. You can run AI locally and measure the energy use yourself. It’s the same as playing a 3d game.

    Edit: It’s literally the same energy use as playing MarioKart on a Switch for 20 minutes.

    https://www.qt.io/blog/examples-of-local-llm-usage

    • queermunist she/her@lemmy.ml
      link
      fedilink
      arrow-up
      1
      ·
      edit-2
      2 days ago

      Carbon emissions don’t just happen at point of use, they have to train these chatbots. That takes billions of kWhs of electricity, there’s a reason they want to build nuclear reactors to power their data centers.

      Then there’s the immense water demands, which can cripple the actual human access to drinking water.

      So, sure, the energy isn’t used when you generate a picture. I misspoke. It was already used, but by creating demand it only incentivizes further training with even higher energy and water demands.

      Also, I noticed you just glossed over the fact that they’re trained on stolen art.

      • Blue_Morpho@lemmy.world
        link
        fedilink
        arrow-up
        1
        ·
        edit-2
        2 days ago

        Gpt 4 had a training cost of $78 million. Gta5 cost $300 million. 4000 developers each with the latest GPU burning hundreds of watts per employee to create the assets. A rough estimate of 750watt pc, 4,000 developers, 8 hour a day, 300 days a year, 5 years = 36 giga watt-hours. That’s the energy to power 3.6 million homes for a year and I’m not even including the HVAC costs of the office space.

        For 1 game.