As climate change increases risks to infrastructure and communities, artificial intelligence is bringing to light new ways to understand hazards, anticipate their effects, and evaluate potential responses. The second installment of the BBISS Insights Series explored these themes, bringing together Georgia Tech researchers in climate science, computational modeling, infrastructure systems, and human-centered design. The discussion examined how AI can help researchers model extreme weather, accelerate complex simulations, guide infrastructure investments, and understand the effects of technological and environmental change on communities.

The session began with an update from Jennifer Chirico, associate vice president of Sustainability, on Georgia Tech’s Climate Action Plan, which helped to contextualize resilience in terms of the campus and community. Chirico also talked about Georgia Tech’s zero waste efforts and the Campus Resilience and Adaptation Framework.

Connecting AI and Physical Systems

Ali Sarhadi, assistant professor in the School of Earth and Atmospheric Sciences and director of the Climate Risk and Extreme Dynamics Lab, researches and models hurricanes, which generate three main hazards: wind, storm surge, and rainfall. What is less commonly understood is that these hazards can be self-reinforcing to cause a phenomenon known as compound flooding. Understanding and predicting how compound flooding occurs requires models that account for complex interactions in coastal and inland environments.

Because historical records provide limited information that is relevant to compound flooding events, Sarhadi uses physics-based models to generate thousands of simulated hurricanes under current and projected climate conditions. Machine learning can connect those simulations with information about buildings and infrastructure, helping translate modeled hazards into estimates of potential damage. AI can also help evaluate possible adaptations by exploring combinations of engineered and nature-based mitigation strategies to see how they fare under an onslaught of digital hurricanes.

Suhas Jain, assistant professor in the George W. Woodruff School of Mechanical Engineering, studies the underlying physics of how fluids flow and mix. Phenomena such as clouds, breaking waves, ocean mixing, wildfire plumes, and ice-ocean interactions have underlying processes that can be extremely computationally demanding to simulate accurately. Jain uses AI to accelerate such simulations by bridging the gap between the physics of what can be modeled at the microscopic scale, and faster, less detailed models.

By rigorously modeling fluid flow and mixing at the micro-scale, he is able to achieve better and faster results than larger models by using machine learning and AI to identify relationships between experimental and simulation data. Jain emphasized that AI can augment physics, but can’t replace it.

From Reactive to Anticipatory Resilience

Iris Tien, professor in the School of Civil and Environmental Engineering, extended the discussion from environmental modeling to infrastructure decisions. Water, power, transportation, communications, and other critical systems are interconnected and increasingly exposed to climate hazards. Yet infrastructure is frequently managed reactively, with repairs and investments occurring after problems emerge.

Tien discussed a more anticipatory approach that considers future hazards and changing community needs before failures occur. In research focused on coastal Georgia, she and her collaborators combined sea-level-rise projections, flood risk calculations at the resolution of the building, development’s impact on water demand, and information about water-distribution networks to identify infrastructure systems that would benefit most from an accelerated maintenance schedule.

A related project revealed that community preference for infrastructure solutions is a critical factor in choosing among proposed construction projects. Tien uses digital twin models to show how different design solutions perform under storm conditions. Because different approaches offer different benefits, the preferred solution can depend on what the community stakeholders value. Informing community members in this way leads to better outcomes financially, functionally, and from a policy perspective.

Keeping People and Places at the Center

Abigail Stengel, assistant professor in the School of Industrial Design and director of the Multisense Studio, raises a different question: What do people need in order to live, work, and participate with their communities inside of landscapes being transformed by AI, manufacturing, and the energy transition? Her presentation considered Georgia’s growing electric vehicle and digital infrastructure landscape, where factories, data centers, transmission infrastructure, farms, watersheds, workplaces, and communities increasingly intersect.

Through a BBISS-supported project, Stengel and her collaborators studied workforce experiences in sectors affected by the electric vehicle transition. Georgia Tech students interviewed workers about their experiences entering, participating in, and advancing within changing workplaces. The preliminary research identified challenges involving training, language, safety, belonging, job demands, and access to opportunity.

The work underscored the importance of understanding technological transitions not only through infrastructure and economic development, but also through the experiences of workers and communities living through those changes.

Resilience as an Interdisciplinary Practice

In all of the presentations, it became clear that AI is being used to accelerate resilience research. Georgia Tech researchers are keeping communities at the center of their work and ensuring that people are always involved in formulating and verifying their research processes and outcomes.

This BBISS Insights Series event showed how the nature of resilience research requires interdisciplinarity.  Shocks to the many systems in which we live our daily lives are difficult to predict and plan for and can come in many forms, both natural and human derived. The researchers who presented their work demonstrate how AI can reinforce and accelerate resilience research, but AI’s greatest value may be to help us understand the connections between people, human-made systems, and natural forces at every scale.

Next in the Insights Series

On Wednesday, Nov. 18, BBISS will hold the next Insights Series at the John Lewis Student Center in the Piedmont Meeting Room. The topic will be “AI for Ecosystem Sensing.” Ethan Pickering, from the University of Georgia, will discuss his lab’s purpose-built AI for biological discovery and crop design. Katherine Graham, from Georgia Tech, will show how her lab uses molecular biology, microbiology, and modeling tools to address issues of water, climate, and health domestically and internationally. Longxiang Li, from Emory University, is developing a satellite-based machine learning model to better understand the potential health effects of exposure to environmental pollutants. Registration is required.