Ganguly Leads New AI for Climate and Sustainability Focus Area

""

Auroop Ganguly, COE distinguished professor of civil and environmental engineering and director of the Sustainability and Data Sciences Laboratory (SDS Lab), is leading a new AI for Climate and Sustainability (AI4CaS) focus area to develop sophisticated new solutions around climate risk, resilience, and sustainability.


Harnessing AI to Turn Complexity Into Planet Saving Innovations

New AI for Climate and Sustainability (AI4CaS) focus area will use research, entrepreneurship, and training to fight climate change.

The Earth’s climate is determined by an intricate web of ecosystems that interact across local and global scales. Turning that complexity into actionable insights, like building coastal resilience or predicting natural disasters, is where human-centered and knowledge-guided artificial intelligence thrives.

The Institute for Experiential AI at Northeastern University hopes to take a leading role in those efforts with its new AI for Climate and Sustainability (AI4CaS) focus area. A joint effort with the Roux Institute at Northeastern University, the AI4CaS team aims to develop sophisticated new solutions around climate risk, resilience, and sustainability, such as the nexus of extreme weather with water, energy, food, ecosystems, and infrastructure. It will accomplish this by combining cutting edge AI and data science tools with fields like physics, chemistry, geoscience, biology, engineering, social sciences, and policy.

AI4CaS is led by Auroop Ganguly, who in addition to being a core faculty member at the institute is a distinguished professor in the Northeastern University College of Engineering and the director of the Sustainability and Data Sciences Laboratory (SDS Lab).

“Traditional tools struggle to work with very large, complex data that interact with each other on multiple scales across space and time,” Ganguly says. “We’re hitting limits in our ability to use this information. That’s where AI and machine learning come in.”

As an example, Ganguly cites recent advances in using generative AI and hybrid-physics AI for short-term precipitation forecasting, a topic on which he conducted his dissertation research 20 years ago.

AI4CaS will address climate and sustainability issues through research, training, and entrepreneurship. Much of that work is already underway. Through the SDS Lab, Ganguly conducts research on some of the world’s most pressing issues with fellow academics, national laboratories, large corporations, and government agencies. AI4CaS plans to expand that work through institute resources like the AI Solutions Hub and Responsible AI Services.

“The world is too disciplinary when the real need is for interdisciplinary solutions,” Ganguly says. “People can talk about convergence research, but there are very few academic, industry, or government labs doing exactly what we do. It’s the combination of Northeastern being an interdisciplinary university and an experiential university. Consider my SDS Lab. As a PI, I have had a non-traditional pathway to academia, preceded by more than a dozen years in the private sector and a government lab. Almost all of my PhD students here are in interdisciplinary engineering and come from areas ranging from computer science to civil engineering, electrical and industrial engineering, machine learning and statistics, and social sciences or policy. The same is true for the SDS Lab and AI4CaS research scientists. That combination just doesn’t exist in other places.”

Another key differentiator for AI4CaS is its human-centric approach to AI, which the Institute for Experiential AI is already well known for.

“We believe in the value of having human decision makers in the loop,” Ganguly says. “On top of that structure, the information we get from AI has to be processed so it’s useful. That’s where interpretability and explainability are key. We are able to bring in various knowledge-guided AI systems, where we have human-centered AI and data-centered AI. In contexts like climate, energy, and water, there are just not that many organizations that can bridge that gap.”

A Multi-pronged Approach

Ganguly says AI4CaS research will focus on both problems and solutions. Problems AI4CaS hopes to address include how to prepare for natural disasters, how to allocate resources like food and water, how to design and operate infrastructures and lifelines, and how to manage and protect migratory populations.

Some of the solutions AI4CaS hopes to provide include more climate resilient critical infrastructure, natural disaster prediction services, and new ways to produce and distribute energy. All of that work will center on the idea of building human-centric AI that is informed by physics, biology, geoscience, as well as engineering and policy principles.

“The way we’ve traditionally handled complexity is to build these physics, geochemical, and techno-social models,” Ganguly says. “Now there are these huge datasets, and machine learning and AI can help us deal with these large data and get us toward solutions in this space in a way that integrates human knowledge, models that encapsulate part of that knowledge, as well as data-driven methods and AI.”

For Ganguly, startups are a natural byproduct of the way he conducts research: in close collaboration with companies and other groups trying to solve real-world problems. Zeus AI, a NASA funded spinout led by former PhD students from Ganguly’s lab, produces high-resolution weather forecasts that account for wind, temperature, humidity, solar irradiance, and more. A startup from the SDS Lab, risQ, modeled the financial risks of climate change in urban regions and translated them into actionable insights for property owners and developers. risQ was acquired by a Fortune 500 company called Intercontinental Exchange, which counts the New York Stock Exchange as a subsidiary.

“You need to always be talking to stakeholders and thinking about real problems and what tools you can bring in,” Ganguly says. “Startups have to be fairly focused. They need to solve specific problems, and these problems come from the research areas we’re focused on.”

The final thrust of AI4CaS is training, which includes educating students but also matching students and postdocs with employment opportunities and upskilling members of partner organizations to leverage AI’s capabilities.

“We’ve heard from partners with long-time employees who are extremely valuable, but may not understand the latest AI or data systems enough to incorporate them into their solutions,” Ganguly says. “We’ve gotten requests for some of those groups to send employees to work with us or take training with EAI and AI4CaS.”

Scaling For Impact

Ganguly’s research affiliations include some of the most prominent national labs in the country and span the globe. He worked at Oracle Corporation for about five years and at Oak Ridge National Laboratory (ORNL) for seven years prior to joining Northeastern. He also currently has a joint affiliation with Pacific Northwest National Laboratory. His research and development activities have been funded by organizations including the US Department of Defense, the US Department of Homeland Security, the US Department of Energy, the National Science Foundation, as well as by private foundations and businesses. In addition, some of his current research is being funded by NASA, and many of his former students have gone on to work at organizations like NASA and ORNL, as well as private companies such as Microsoft, Tokio Marine, and AIR Worldwide / Verisk.

Ganguly has also worked with private companies in the insurance, finance, and energy sectors. He believes that work will provide the group with a unique perspective.

“Businesses have to be focused on the relatively short-term bottom line, and national labs are mission driven, so this ability to balance between breadth and depth — between short-term versus long-term solutions — makes us very well positioned,” Ganguly explains.


Source: The Institute of Experiential AI

Related Faculty: Auroop R. Ganguly

Related Departments:Civil & Environmental Engineering