AIM Seed Grants Support 11 AI Research Projects

From tracking the impact of artificial intelligence data centers to investigating how responsible AI technology could help alleviate homelessness or support students in the classroom, 11 new seed grants will fund projects by faculty from across the University of Maryland who are advancing AI innovation in ways that uplift society.

The grants were selected through the Artificial Intelligence Interdisciplinary Institute at Maryland's (AIM) 2026 Research Seed Award Program, which supports innovative science and scholarship advancing foundational AI research as well as AIM's core focus areas of accessibility, justice, sustainability and learning. 

Together, the funded projects explore how emerging technologies can address complex societal challenges and generate meaningful public impact:

From Singular Trajectories to Shared Insights: Individualized and Generalizable AI Models of Social and Agency Development in Autism
Led by Clark Leadership Chair Professor Fengfeng Ke in the Department of Teaching and Learning, Policy and Leadership and Professor Yi Ting Huang of the Department of Hearing and Speech Sciences, the project will develop a dual-scale, competency-based modeling framework to trace the development of social-communicative and relational agency capacities in autistic adolescents while integrating practitioner expertise within computational modeling workflows. In partnership with three autism-serving organizations, the team will develop and refine analytic tools that combine insights from behavior analysts, therapists and educators with data-driven modeling.

From Bench to Bedside and Back Again: Co-Designing AI-Supported Clinical Interventions for Aphasia
Guided by clinical and human-computer interaction approaches, the project led by Assistant Professor Stephanie Valencia-Valencia of the College of Information and Professor Yasmeen Faroqi-Shah of the Department of Hearing and Speech Sciences will iteratively co-design AI-supported interventions for people with aphasia, as well as implement longitudinal personalization and evaluate the tools. Concurrently, it will uncover accessible interaction mechanisms, fine-tuned models and lucid explanations for AI support.

Investigating the Cognitive Underpinnings of Social Biases in LLMs
Led by computer science Assistant Professor Rachel Rudinger, linguistics Professor Naomi Feldman and computer science Assistant Professor Sarah Wiegreffe, the project uses concepts from cognitive psychology, investigating deviations from rationality, or cognitive biases, in large language models (LLMs). The project will also investigate the extent to which social and cognitive biases in LLMs share underlying mechanisms.

AFO Mapping and Spatially-Explicit Nutrient Transfer Optimization
Geographical sciences Assistant Professor Catherine Nakalembe and environmental science and technology Professor Stephanie Lansing are leading research to optimize the placement of waste-to-resource technologies across Maryland's agricultural landscape by resolving critical data gaps in animal feeding operation (AFO) mapping and crop nutrient demand. The project employs a geospatial AI approach that integrates deep learning with multisource satellite imagery for high-resolution AFO detection and nutrient supply and nutrient balance mapping, as well as geospatial optimization. 

AI-Driven Identification of Homelessness and Unstable Housing in Healthcare Administrative Data
Health policy and management Assistant Professor Dahai Yue is developing and refining a first-of-its-kind artificial intelligence algorithm to identify and characterize homelessness using California statewide Medicaid administrative data. The resulting scalable tool will enable policymakers and service providers to target prevention, outreach and care coordination more effectively.

Democratizing Resource Allocation: An Accessible AI Optimization Platform for Real-Time Sustainable Decision Making
Haizhao Yang, an associate professor in the Department of Mathematics and the Department of Computer Science, is leading a research team that will develop a real-time, cloud-based AI optimization platform that enables users to solve complex resource-allocation and decision-making problems through natural language or voice input. By removing barriers associated with specialized expertise, coding skills and availability of computing hardware, the project seeks to democratize access to advanced optimization tools. 

Discovering Changing Values in the AI Workforce
Led by College of Information Professor Katie Shilton, the project explores how artificial intelligence is changing workplaces and practices, with the workers producing AI technologies among the most immediately impacted. The project seeks to identify skills needed to support future AI workers and potential value barriers to responsible AI development.

REPS-AI: Laying the Intellectual Foundations for Supporting All Students in Using GenAI Tools in Responsible, Effective, and Pedagogically Sound Ways in Computational Coursework
Associate Professor David Weintrop, who has a joint appointment in the College of Information  and the Department of Teaching and Learning, Policy and Leadership, is leading a project that seeks to identify the foundational concepts and practices that support undergraduate students in using generative AI tools responsibly, effectively and in pedagogically sound ways within computational coursework. The project will develop a taxonomy of productive GenAI concepts and practices that can serve as a foundation for courses, workshops and student support resources.

AI Infrastructure and Energy Justice: How Data Centers Reshape Community Vulnerability to Power Outages and Health Impact
Professor Yueming "Lucy" Qiu, the Roy F. Weston Chair in Natural Economics and associate dean for research and faculty affairs in the School of Public Policy, is leading a project that investigates how AI data centers affect nearby communities. Using a nationwide county-level panel dataset, the research will assess how data center presence and expansion influences outage frequency, duration and recovery disparities. The project will also use machine-learning methods to estimate backup generator deployment and associated localized air pollution and health costs, informing equitable policies for data center siting, utility restoration protocols and emission controls.

A User Needs Analysis for AI-Enabled Facial Blurring in Protest Photos, Videos, and Live Streams
College of Information Professor Jennifer Golbeck is leading research that will analyze users’ needs regarding automated, AI-enabled face-blurring technology in photos, videos and live streams of protests and demonstrations. By understanding real-world conditions along with concerns and goals of citizen users, the project seeks to explore how AI technologies can be leveraged to protect individuals from surveillance.

Reducing AI Overreliance in Individual-Centric Domains
Aneesh Rai, an assistant professor in the Department of Management and Organization, is leading a project that examines how to reduce overreliance on AI in individual-centric domains, or creative and cognitive contexts where the long-term benefits of individual effort match or exceed short-term gains from delegating to AI. Using surveys, vignette studies, behavioral experiments and a field experiment, the project seeks to generate actionable insights for individuals, organizations and policymakers seeking to promote responsible AI use.