The idea that personal LLM use is environmentally irresponsible shows up *a lot* in many of the online spaces I frequent. I've [touched on my doubts around this](https://simonwillison.net/2024/Dec/31/llms-in-2024/#the-environmental-impact-got-better) in the …
  • Sneezycat@sopuli.xyzEnglish
    7·
    1 天前

    If you want to prompt ChatGPT 40 times, you can just stop your shower 1 second early. […]

    Notably, Andy’s calculations here are all based on the widely circulated higher-end estimate that each ChatGPT prompt uses 3 Wh of energy.

    So, 1s of shower is supposed to be 40*3Wh, or 120Wh? That’s, what, a 432000W heater? I’m not running a small nuclear power plant to shower in the morning.

    • cx40@programming.devEnglish
      3·
      1 天前
      • Specific heat capacity of water: 4.184J/(gC)
      • Average shower temperature: 37C
      • Cold tap water temperature: T
      • Shower head output: 2gallon/min = 7.57L/60s = 126ml/s = 126g/s

      1s * (126g/s) * (4.184J/(gC)) * (37C - T) = 524J/C * (37C - T)

      120Wh = 432000J

      T = -787C

      So I guess the math checks out if your city’s water supply temperature is way below absolute zero.

      Scratch that, let’s assume the water actually flows, so T=0C. What water throughout do we need to achieve this level of power consumption?

      1s * (X g/s) * (4.184J/(gC)) * (37C) = 432000J

      X = 2791g/s = 44gpm

      You would have to be showering with 17 showerheads simultaneously using showerheads that are rated at the highest legal flow rates in the US (2.5gpm).