Rolnick et al. (2022) assembled a comprehensive survey with contributions from leading AI and climate researchers, identifying 13 key areas where machine learning can meaningfully contribute to climate change mitigation and adaptation:
- Electricity systems: AI-optimized power grids can reduce energy waste by 10-15%
- Transportation: Route optimization, autonomous vehicles, and demand prediction
- Buildings: Smart HVAC systems that reduce energy consumption by 20-30%
- Industry: Process optimization, supply chain efficiency, materials science
- Agriculture: Precision farming, crop yield prediction, reducing food waste
- Forests & Land Use: Satellite-based deforestation monitoring, carbon stock estimation
- Climate prediction: Improved weather and climate models
The Net Question
While AI consumes significant energy (see Strubell 2019, Patterson 2021), this paper argues the net impact of well-applied AI on climate is overwhelmingly positive — the energy savings enabled by AI far exceed its own consumption.
Source
Rolnick, D. et al. (2022). ACM Computing Surveys, 55(2), 1-96.