Dr. Gengchen Mai has been awarded a $300,000 Methodology, Measurement, and Statistics Research Grant from the National Science Foundation. This grant will fund Mai’s research project entitled “A Statistics-Based Geographic Bias Quantification and Debiasing Framework for GeoAI and Foundation Models.” Dr. Gengchen Mai is an Assistant Professor at The University of Texas Department of Geography and the Environment.
Pioneering studies show that AI models, especially foundation models, can amplify biases embedded in the data, leading to severe societal outcomes. Although numerous efforts have been made to quantify and mitigate bias in AI models, geographic bias receives much less attention which presents unique challenges. With the NSF grant funding, Dr. Mai and his collaborators aim to develop a statistics-informed framework to measure and mitigate geographic bias for both task-specific geospatial AI models and foundation models.
Dr. Mai believes geographic bias quantification and debiasing is an important but usually overlooked challenge in trustworthy AI. The proposed framework represents a pioneering effort to bridge the gap between AI and societal values, with broad implications for the ethical development of AI across multiple domains.
This project has been made possible by support from many. Dr. Mai thanks his co-Principal Investigator, Dr. Ninghao Liu from the University of Georgia. He also would like to thank his collaborators, Dr. Guofeng Cao from University of Colorado Boulder and Dr. Ni Lao from Google DeepMind. He also extends his thanks to Dr. Jennifer A. Miller, the Chair of the Department of Geography and the Environment, and to Teal Reid, the Executive Assistant and Department Co-Manager in the Department of Geography and the Environment. Lastly, Mai extends his thanks to Jeff Meserve, Assistant Director of Research Development in the Humanities Institute.
