Inspiration
As generative AI becomes an integral part of our lives, we must confront its environmental impact. Many of us are oblivious to the big-picture effects of our daily actions. Inspired by existing calculators like WWF’s Footprint Calculator, we wanted to bridge the gap between sustainability and AI.
What it does
EnerGPT is a GoCogle Chrome extension designed to calculate and visualize the electricity and water consumption associated with your ChatGPT usage. By storing and displaying daily and weekly data, EnerGPT offers insights into the hidden energy costs of generative AI. The extension translates energy consumption into accessible metrics, such as the electricity needed to charge your phone or the water required to grow blueberries, helping users grasp the true environmental footprint of their AI interactions.
How we built it
Our extension has three core components: collecting user activity, translating it into energy consumption, and visualizing the data. We started with research on statistics and Chrome Extension development. Armed with concrete information and an understanding of the extension pipeline, we began solving how we would be collecting and storing data. Using JavaScript, we tracked user prompts by listening to calls to the ChatGPT backend API, and persisted daily and weekly data using Chrome Storage. We then converted this data into meaningful energy consumption metrics. To display the results, we utilized Chart.js to create dynamic bar graphs for weekly and daily usage. Finally, with HTML and CSS, we polished the project with a user-friendly interface.
Challenges we ran into
As a team of beginners, we had many challenges along the way. Our limited experience with GitHub collaboration posed frequent obstacles—merging conflicts and repository management. Our first ideas of webscraping in order to track user activity turned out to be a naive approach, which depleted much our time at the beginning. Then Chrome storage also posed to be a big learning curve for many of us as we faced many obstacles in figuring out how to store our data both daily and weekly.
Accomplishments that we're proud of
Despite these challenges, we maintained productivity and positivity throughout. We were able to collaborate very effectively with mindful and efficient communication. Despite the limited time, we managed to stay on track and were never rushing against the clock, keeping ourselves calm as we debugged and developed our program. Most of all, as a team of first-timers, our biggest accomplishment was bringing an idea from research to product while having fun.
What we learned
A lot of the challenges we faced turned out to be great learning experiences. Every team member was able to walk out of this event feeling more confident about their skills with GitHub, JavaScript, HTML and CSS, and Chrome Development Tools. We were able to learn efficient collaboration and communication on top of technical skills.
What's next for EnerGPT
As research on generative AI’s energy impact gains traction, we plan to integrate more detailed, proximity-based metrics—like carbon emissions and electronic waste—from data centers. We also intend to expand our scope beyond ChatGPT to include AI platforms such as Gemini, Copilot, and Claude, and account for various AI interactions (e.g., image creation and API calls) that might require new energy consumption metrics.
Log in or sign up for Devpost to join the conversation.