JANA THOMPSON
Sustainable AI
Architectural Practices for Inspiration
The following are from architectural projects that I have found inspiring over the last couple of years in my work.
An MIT Media Lab professor and architect with her own firm Oxman, her research looks at, in her own words, "accelerating systems-level change through nature-centric design". Her explorations of sustainable materials with technology and spatial playfulness are both delightful and deeply provocative.
Société d'Objets Cartographiques
Co-founded by French landscape architect Alexandra Arènes in 2016, SOC is a think tank focusing on earth political design, drawing on scientific and public enquiries, and producing workshops and exhibitions.
Material Cultures is a UK-based architecture consultancy focused "on design, material research, and high level strategic thinking to make meaningful progress towards a post-carbon built environment."
Carbon Measurement Tools For Developers
The following are links for measuring carbon output of cloud usage for developers
A tool developed for looking at the carbon footprint of public cloud usage. It works through a command line interface, so ideal for a developer or to be developed into an app.
Python package for developers that can be installed using pip and used easily within a development pipeline. Can produce visualizations to help easily explain impacts for a project.
This tool is designed more specifically for AI researchers to calculate the carbon cost of projects and add to the publications, but has a web interface for easier usage
Stories (Some With Resources Listed)
The following are stories or links with information that some may find relevant for meeting sustainability needs
Enterprise Solutions for Carbon Measurement
I have never personally used these or looked at these in depth, but for an enterprise looking at meeting sustainability goals, these might be an excellent resource.
Papers
A list of papers on the topic of measuring ecological impact for AI projects. Note: most of these are on the more technical side and geared for AI researchers and developers
Dodge et al. 2022. Measuring the carbon intensity of AI in cloud instances. Presented at FAacT 2022: https://facctconference.org/static/pdfs_2022/facct22-145.pdf
Ligozat et. al. 2022. Unraveling the hidden environmental impacts of AI solutions for environmental life cycle assessment of AI solutions https://arxiv.org/pdf/2110.11822.pdf
Woolf Anthony et al. 2020. Carbontracker: Tracking and predicting the carbon footprint of training deep learning models. https://arxiv.org/pdf/2007.03051.pdf