AI Institute for Climate Resilience (ACRE)

Praveen Kumar and Alexandre Tartakovsky (Civil & Environmental Eng)
Kyratso Karahalios, Nan Jiang, Ruta Mehta, and Arindam Banerjee (Computer Science)
Jeff Shamma (Industrial & Enterprise Systems Eng)
Shaowen Wang (Geography & GIS – LAS)
Sandy Dall’Erba (Agr & Consumer Economics – ACES)
Volodymyr Kindratenko and Ashish Sharma (OVCRI Institutes)

Research Problem

Non-stationarity results in historically rare events becoming more frequent due to climate change, resulting in compounding and cascading unforeseen failures across a range of coupled human-natural systems. Despite the urgency, action on climate and weather extremes is hindered by many barriers, exacerbated by knowledge gaps in causal understanding across disparate systems and cognitive dissonance in decision-making across scales.

ACRE Vision

Develop foundational AI advances that provide a new paradigm for using systems that learn and support causal reasoning to enable evidence-based and informed choices anchored in social and environmental justice for decision-making.

Larger Impact

The foundational advances in AI planned in ACRE aspire to build a general framework for decision-making for such societal challenges and improve all critical constituent aspects, including improving and validating mechanistic models; casual models to enable informed and reasoned decision-making supported with counterfactual ‘what-if’ scenario analysis; characterization and assessment of the fairness of decisions; characterization and modeling of tail behavior (extreme events) and associated uncertainties; and core advances in decision making formulations based on reinforcement learning, control theory, and contextual bandits, among others.