Skip to content
Awesome AI for Industrial Ecology

OverviewA proposal for the AI4IE working group

Awesome AI for Industrial Ecology

What AI already does in industrial ecology (IE) research, the papers behind it, the gaps still open, and how to try it without getting burned.

Open the applications mapHow to contribute
LCA & footprinting → LLMs & agents: 8LCA & footprinting → NLP & information extraction: 3LCA & footprinting → Tabular ML & regression: 7MFA & stocks and flows → Computer vision: 5MFA & stocks and flows → Detection & segmentation: 2MFA & stocks and flows → Remote sensing: 5MFA & stocks and flows → Tabular ML & regression: 3Waste, EoL & circularity → NLP & information extraction: 1Waste, EoL & circularity → Computer vision: 11Waste, EoL & circularity → Detection & segmentation: 4Waste, EoL & circularity → Tabular ML & regression: 2Waste, EoL & circularity → Foundation models & transformers: 2Urban mining & secondary resources → Computer vision: 4Urban mining & secondary resources → Detection & segmentation: 1Urban mining & secondary resources → Remote sensing: 1Urban mining & secondary resources → Tabular ML & regression: 1LCA 14MFA 10Waste, EoL 13Urban mining 4LLMs 8NLP 4Computer vision 15Detection 5Remote sensing 5Tabular ML 10Foundation models 2
Fig. 1 What the 34 AI-in-IE papers apply, from IE field to AI technique. Numbers count papers; band width counts field-technique pairings (60), since one paper can sit in several bands. Follow a label or band to read the papers. Wanted No paper yet applies Depth & 3D geometry, Generative models, Anomaly & condition detection to an IE field.Source: src/data/literature.yaml

Industrial ecology studies how materials and energy move through the economy: what products are made of, how long they last, where they end up, and what that costs the environment. AI can help with a lot of that work, and it can also produce confident nonsense. This guide sorts the two and points to the papers behind each verdict.

It’s community-curated, started in the AI4IE working group: IE researchers scattered across institutes who meet on Discord. Anyone can contribute.

How it fits together

Section titled “How it fits together”
  1. Step 1See what works todayMap of AI applications7 computer-vision applications at end-of-life, each with a maturity call
  2. Step 2Read the evidenceLiterature & resources11 papers tagged waste & EoL × computer vision
  3. Step 3Try itDemosZero-shot recognition of your own photo, in this browser tab
Fig. 2 One path through the site, worked for waste & EoL × computer vision. Every field and technique follows the same three steps, because one tag vocabulary links the map, the literature and the demos.

Who this is for

Section titled “Who this is for”
  • A life cycle assessment (LCA) or material flow analysis (MFA) researcher. Start with what works in your field, then the ML-reproducibility checklist before your first model. Contribute: examples of AI in your field.
  • A student. Start with getting started and the ML landmarks reading list. Contribute: the pitfall you wish someone had warned you about.
  • A research software engineer (RSE) or tool builder. Start with tools & hardware and the how-to-add-a-demo guide. Contribute: a demo of your method, embedded or in-browser.
  • Curious, or anyone else. Start with Fig. 2 above, or try the photo demo. Contribute: an answer to any Wanted mark on the site (each gap on the map links to a short form), or a pull request for any other change.

Contributing

Section titled “Contributing”

No pull request needed: suggest a paper, map an application, suggest a tool or suggest anything else. Each is a short form on GitHub, and a maintainer turns it into an entry. The forms need a free GitHub account. Without one, post your suggestion in the Discord.

To make the change yourself, the contributing guide explains the format and where each type of entry goes. The data lives in two YAML files, and every page has an edit link.

Content CC BY 4.0 · Code MIT