Devlog #2: AI Footprint Detection

What I Have So Far
The extension now accurately detects and displays the stopwatch when the user is on an AI website (specifically ChatGPT, Claude, and Gemini so far). I had some difficulty with this due to certain variables being set at weird times and slight inconsistencies with event listeners, so the time wouldn’t store correctly or keep resetting. Along with fixing that, I added the ability for the stopwatch time to be stored within the user’s chrome storage and a daily reset. Finally, I cleaned up the pop-up appearance.

What I Need To Work On
My next step is user personalization, both in display and the actual calculation, for which certain values are user-dependent (location, device, etc.). I have already begun to draft the location part but the code is commented out right now as to not break the emissions count right now. The ultimate goal is detecting key strokes for a more accurate measure of time using AI, but that’s for the future.

Devlog #1: AI Footprint (tentative name)

I finally have some semblance of a chrome extension to submit evidence of! This is my first time working on an extension so majority of this time was spent experimenting and figuring out how to structure the program.

The Concept
With the increased usage of AI, I believed it was necessary to bring to light its environmental impacts. Most people think of its costs in general terms: takes up land, a water guzzler, produces carbon dioxide. This chrome extension displays real-time data for people who like the see the actual numbers.

What I Have So Far
The extension opens successfully, the popup displays as intended, and the stopwatch and carbon emissions tracker run when the user is on the Claude website.

What I Need To Work On
At the moment, the extension only detects Claude. I have yet to figure out why it does not work on ChatGPT or Gemini (my two other test subjects) but that is what I’ll focus on next. In addition, it assumes average values for power draw and grid carbon density among others which all depend on geographical location and device processor, which I intend to turn into user inputs at a later point.