GPT Footprint Dashboard

GPTfootprint estimates the environmental footprint of ChatGPT responses by tracking the number of output tokens in a response. (Tokens are sort of like the word count.) Using the user's coarse location, based on the country, state (if in the United States), or county (if in Pennsylvania), the extension estimates carbon, water, and energy usage associated with that activity. The extension only collects the number of output tokens and coarse location data. It does not collect or store the contents of user prompts or responses. Usage data is sent to gptfootprint.cs.haverford.edu so users can view total carbon, water, and energy usage across all users.

Total Users:

Tokens KWh CO2e (g) Miles Driven Water Consumption (mL) Water Bottles (500 mL)
Total across all users
Average Per User Usage Total
Predicted Haverford Annual Academic Usage

Adjusted Water Impact (AWI) by Watershed

PFAF ID Short AWI Long AWI (0.1) Long AWI (0.01) Locations