I'm putting together a small art project, aiming to highlight both local and global environmental considerations. The idea is that every hour the system generates a sentence based on current air quality measurements, e.g. "the air quality is good", this then gets sent to a neural network in the cloud that generates a unique image based on that sentence. The image then gets sent back to the device to be displayed on a 7 colour eInk display. I've made a small prototype of that section of the project already (without air quality sensors yet), here you can see it displaying an image for the sentence "the air quality is very poor":
In addition to sending back the image, I plan on making the cloud server also send back the estimated CO2e required to generate it. This can then be displayed alongside the image, with some illustrative comparisons, e.g. "This image required 10g of CO2e to generate, that's the equivalent of driving 400 metres in an average car."
Finally, the viewer is asked if they think that particular generated image was worth the expenditure of CO2 it required, with "Yes" and "No" buttons available for them to indicate their opinion. Statistics for each image's popularity will then be displayed on the project's website.
I'm looking forward to seeing what everyone else gets up to :)