Priya Donti, now a faculty member at MIT, blends her passion for equity with machine learning to tackle one of the toughest challenges in climate action: managing power grids with growing renewable energy, says MIT News. Her motivation partly stems from witnessing global inequalities during childhood visits to India, and later from teaching that tied climate to justice. What attracted her to her current path was seeing how AI could unlock new ways to balance variable solar and wind generation.
At MIT, Donti develops algorithms that don’t just forecast power output from renewables, but actively incorporate the physics and constraints of real grids. Her models aim to help operators make decisions that are both cost-efficient and physically reliable. In benchmark tests, one algorithm she developed runs 10 times faster and much more economically than existing methods, though it’s not yet in full deployment.
A major obstacle in this domain is data privacy. Much of the operational data needed to train models is private or sensitive. To address this, Donti’s group builds “synthetic data” and shares benchmarks that simulate grid behavior, enabling research without compromising proprietary systems.
Her reach goes beyond algorithms. She co-founded Climate Change AI, a nonprofit that connects technologists, domain experts, policymakers, and practitioners who care about climate and technology. At MIT, Donti will co-teach a course called AI for Climate Action, joining forces with colleagues focusing on ecosystems and Earth sciences.
Priya Donti’s work shows that AI’s potential in climate isn’t purely theoretical; it’s about building tools that respect real infrastructure, human context, and the urgent need to shift energy systems.