In an ever-shifting landscape of (often competing) research about the environmental impacts of AI, keeping up to date is tough. It also gives opportunities to (re)think about what and how we communicate the ethics of AI to students.

And so… v6 of the Impacts Estimator is now live: https://ifyouuseme.ai/environmental-impacts

This has used an updated literature review and Claude Fable. On of my design parameters was to remove any JS or other dependencies from the app, so it is all pure HTML. I also wanted to make it more information-dense, yet visual. This is a complex topic; the linear view of the old version didn’t promote interconnectedness very well.

Update 1: Interactive Explainer

I wanted to do this, so I did it. The Introduction tab is now replaced by an interactive explainer. Users can click about to see summaries of different parts of the system, their impacts, some sources and case studies (which previously had been buried layers-deep in the Lessons tab).

Update 2: New Estimator

Based on updated data (always a best-guess, but drawn from a lot of sources), I wanted to put the parameters next to the “live” impacts and remove the need to “calculate” and wait. I also wanted to lift some example equivalencies from their old home (buried in the lessons tab) and into the app itself, for a more holistic view of environmental footprints. Now within one window, users can see and manipulate many variables.

I removed icons and other space-stealers, to create visual but data-dense view, to generate as many inquiry questions as possible. When you see the screenshot above, how many things could you ask a question about?

I also wanted to try a different type of visualisation, which adapts with the inputs. In this one, the blue circles grow with input.

For transparency and learning, I wanted to keep all of these elements accessible from the same view. They expand with a lot of detail, perhaps useful for environmental science classes:

Other updates are minor and/or cosmetic. Some tweaks to the lesson plans, some broken links fixed and a revised Research list.

Fun fact – since using AI for making apps, I’ve fallen in love with HTML all over again. The Estimator app is only 47kb in total: quick to load and packed with features. I’m determined to avoid JS and other dependencies from now on.