Cross-departmental faculty collaboration with CBI Fellow

Introducing yourselves and your research interests.

My name is Pingbo Tang. I'm an associate professor in the Civil and Environmental Engineering Department at Carnegie Mellon University. My research focuses on the intersection of infrastructure operation, human factors engineering, and climate adaptation. I work to integrate infrastructure engineering knowledge with human-centric working processes, enabling human and artificial intelligence collaboration to enhance the resilience and adaptability of our civil infrastructure, especially in the face of heavy rainfall and large-scale climate-related disasters. My background as an engineer extends into psychology, social science, and human factors.

I am joined by Cleotilde (Coty) Gonzalez, a research professor of Decision Sciences in the Department of Social and Decision Sciences at Carnegie Mellon University. She is the founding director of the Dynamic Decision Making Laboratory (DDMLab) and the research co-director of the NSF AI Institute for Societal Decision Making (AI-SDM). Gonzalez's research focuses on dynamic decision-making, decisions from experience, and cognitive modeling. She developed the instance-based learning theory (IBLT), which has been applied to various domains, including cybersecurity and human-machine teaming.

Working together with CBI fellow Jinfeng Lou has been a productive cross-departmental endeavor, enabled by the Carnegie Bosch Institute.

Introducing your shared work and history of collaboration.

I collaborate extensively with researchers across diverse fields, blending engineering and human systems. I work with photogrammetrists and remote sensing researchers to address large-scale problems like city navigation systems and 3D reconstruction of building sites to understand the spatial relationship between people and their environment. I’ve also collaborated with human systems engineers like professor Nancy Cook at Arizona State University and Ron Boring, Ph.D. at Idaho National Lab, focusing on human-computer collaboration in complex tasks like nuclear plant and airport operations.

Additionally, I work with design researchers in mechanical engineering to explore human-AI collaboration in complex planning and design problems. My interdisciplinary approach began with my structural engineering background and evolved to include management science, construction research, psychology, and public policy. Recently, I began collaborating with an education researcher to enhance learning through machine learning, furthering the integration of engineering, social science, psychology, and education research to build a strong workforce, and resilient cities and communities in a climate-uncertain future.

Gonzalez's work in dynamic decision-making and cognitive modeling complements these interdisciplinary efforts. Her development of the instance-based learning theory (IBLT) provides a framework for understanding how individuals make decisions based on past experiences, which is crucial for designing systems that enhance human-AI collaboration in complex environments.

What about the CBI fellowship drew your attention?

As a domain scientist, I seek the right tools to address civil engineering problems, so the CBI fellowship attracted me because of its focus on integrating AI with domain-specific challenges that demand automated technology solutions. Gonzalez, also a domain scientist, focuses on understanding human learning processes, particularly through instance-based learning methods, highlighting differences in human and machine learning. We aim to identify how AI and human expertise can converge to solve complex issues like urban flooding and subway system resilience in abnormal scenarios, utilizing domain expertise for both AI and human solutions. 

Introducing the fellowship research topic/problem:

This research focuses on creating human-AI collaboration protocols to optimize human and computational intelligence convergence in decision-making with large configuration and solution spaces with uncertainties and missing data. Specifically, we examine urban flooding events, such as the 2021 heavy rainfall in Zhengzhou, China, which necessitated computational data and models to guide decisions on managing subway, pumping, and evacuation systems. We analyze both short-term, real-time responses—such as sensor data utilization and human communication for immediate action—and long-term infrastructure planning and city management to improve resilience against heavy rainfall.

In real-time scenarios, we explore how data can direct evacuation priorities and pump activation to mitigate rainfall impacts, aiming for controlled and gradual failure of subway networks to avoid sudden collapses. We also study human decision-making in simulation environments to understand how human intelligence complements AI in varying scenarios. Ultimately, our goal is to identify fundamental features across global subway network systems that aid operational resilience. Our work will address data shortages, and leverage human expertise to improve decision-making in both short- and long-term scenarios, with attention to organizational and human factors.

Gonzalez's expertise in dynamic decision-making and cognitive modeling, particularly through the development of IBLT, provides a theoretical foundation for understanding how individuals and systems can adapt to changing environments. This is essential for designing protocols that enhance human-AI collaboration in managing urban flooding and related challenges.

Why is this topic important, pressing, or exciting?

This topic is crucial due to the increasing impact of climate change on urban infrastructure, especially the risk of urban flooding in subway systems. Recent events, like hurricanes and flooding in North Carolina, Florida, and globally, have underscored these challenges. Subway flooding disrupts not only transportation but also public health by mixing wastewater and drinking water, posing severe health risks when treatment capacity is insufficient. The research focuses on designing resilient cities and critical evacuation routes within subways, making it an essential area of study.

Scientifically, I'm excited about the complementary nature of human and artificial intelligence. We aim to integrate domain-specific knowledge and tacit decision-making insights into seamless data-driven and symbolic reasoning approaches. Key questions include when human decision-making outperforms AI, how to characterize these tasks, and the situations where human and algorithmic collaboration best reduces risks. By understanding these task execution dynamics, we hope to enhance decision-making resilience, allowing cities to manage failures gradually and gracefully rather than causing community-wide disruption.

Gonzalez's research in dynamic decision-making and cognitive modeling offers valuable insights into how humans learn and adapt in complex environments. Her work on IBLT helps us understand the conditions under which human decision-making excels and how AI can be designed to complement human strengths, making our cities more resilient to climate-related challenges. 

What makes CBI fellow Jinfeng Lou the right person for this topic?

Jinfeng brings a unique perspective. Hailing from Zhengzhou, China—a city recently impacted by unprecedented rainfall—he has a personal understanding of how climate change affects urban communities. His background in information technology and geotechnical engineering, combined with civil engineering, provides a solid foundation for addressing subway flooding and underground infrastructure challenges. His experience spans systems engineering, modular construction, information systems, and decision-making, making him well-suited to explore how AI, information systems, and human capabilities can collaborate to reduce flood-related impacts in urban areas.