intuition calibration

scale intuition for ai environmental impact

this tool helps build intuition for the scale of environmental impact by comparing ai usage (llm inference) with everyday activities. results are estimates based on median values.

1000
estimate uncertainty:

equivalents

🚗
0 km driven ICE passenger car
0 cups of coffee black, no milk
🍔
0 beef burgers single patty

estimated physical impact

electricity 0 Wh
carbon 0 g CO₂e
water 0 mL
disclaimer: all values are estimates. actual environmental impact varies by model size, infrastructure, energy grid, agricultural practices, and geography.

sources

  • llm inference: google cloud (2024), arxiv (2025) "how hungry is ai?"
  • driving: us epa, our world in data
  • food: water footprint network, fao, our world in data

← back to home