The Penn State Center for Socially Responsible Artificial Intelligence (CSRAI) has announced the results of its most recent seed-funding competition. The center awarded more than $152,000 to seven interdisciplinary research projects representing six colleges.
Each proposal was evaluated by peers for its connection to the center’s mission, intellectual merit and potential for securing external funding. The awards will support the formation of interdisciplinary research teams and early-stage projects that demonstrate strong potential to obtain external funding. Projects are expected to start spring 2026 and last one to two years.
“We had another record-setting year in terms of the volume of submissions, making our seed grant program extremely competitive,” said S. Shyam Sundar, CSRAI director and James P. Jimirro Professor of Media Effects in the Donald P. Bellisario College of Communications.
The winning proposals came from a wide range of collaborations involving faculty from a variety of disciplines in science, engineering, information science and technology, health and human development, liberal arts and communication. They were all evaluated positively through a blind peer review process, not only for advancing AI, but also for their social implications, including the extent to which they are responsive to societal needs and their potential for social good.
“Human-centered AI was the winner this year,” said Sundar. "Whether the proposal is to plan for public transportation or urban disasters, or to improve the effectiveness of parent-focused psychotherapy or enhance language learning by non-speaking children, the projects all focus on aligning AI with human social values."
Synopses of the awarded proposals are shared below, with additional information available on the center’s website.
“AI Augmented Urban Resilience Agents with Model-Based Behavioral Priors and Theory-Driven Adaptation for Urban Disasters”
This project aims to create a system that uses advanced AI models combined with real-world data and psychological insights to simulate how people respond and adapt during disasters. The agents will blend realistic data with behavioral science to produce clear, interpretable simulations that improve urban emergency planning, risk communication, and resilience in different disaster scenarios and urban settings.
“Ambiguities and Tensions in Trans Crowdfunding: Viral Visibility Under Surveillance"
This project explores how trans and gender non-conforming (TGN) individuals use online crowdfunding to relocate to safer regions while facing increasing scrutiny and surveillance. Using interviews and discourse analysis, the research examines how platform systems and AI-driven oversight shape the opportunities and risks experienced by TGN crowdfunders, aiming to provide new insight into survival strategies, care practices, and support needs for TGN communities.
-
Kelley Cotter, College of Information Sciences and Technology
-
Andrea Miller, Donald P. Bellisario College of Communications and College of the Liberal Arts
-
Sara Liao, Donald P. Bellisario College of Communications and College of the Liberal Arts
-
Cora Butcher-Spellman, Donald P. Bellisario College of Communications and College of the Liberal Arts
-
Benji Davis, College of Information Sciences and Technology
-
Hil Malatino, College of the Liberal Arts
-
Lili Dudas, College of Information Sciences and Technology
-
Salah Seoudi, University Libraries
“Expanding Empathic AI 2026 – A Research Roadmap”
This project proposes a cross-disciplinary conference focused on advancing understanding of empathy in human–AI interactions. Bringing together scholars from philosophy, the social sciences, engineering, and computer science, the event will examine both the science and ethics of extending empathy to AI systems and interpreting empathy expressed by LLMs and robots. The conference will feature talks, panels, early-career flash sessions, and collaborative workgroups designed to spark new partnerships and produce a shared research roadmap.
-
Christopher Daryl Cameron, College of the Liberal Arts
“Human-Centered Artificial Intelligence for Public Transportation Services”
This project will develop a new AI approach for transit planning that reflects the priorities of community members and transit workers and better aligns with local needs. By combining community engagement with machine learning, the team will test a model that learns and incorporates human preferences into advanced planning tools.
-
Aron Laszka, College of Information Sciences and Technology
-
Yiqi Zhang, College of Engineering
“Leveraging AI to improve therapeutic outcomes in parent-focused psychotherapy”
This project proposes developing a natural language system that offers parents a secure, scalable way to practice relationship-building and behavior-management skills between therapy sessions. The proposed system will provide immediate, iterative feedback to improve skill uptake and address a major implementation gap in current treatment models.
-
Cynthia Huang-Pollock, College of the Liberal Arts
-
Roger Beaty, College of the Liberal Arts
-
Kenneth Huang, College of Information Sciences and Technology
“Leveraging artificial intelligence to enhance language learning by non-speaking children with disabilities who use speech generating technologies”
This project aims to improve speech-generating technologies so non-speaking children with communication disabilities can better develop language skills. This pilot will create and evaluate child and adult language models tailored to children’s language levels and conversational contexts, and design prototype interfaces that provide subtle, real-time scaffolding prompts to both children and their communication partners.
-
Michael Clarke, College of Health and Human Development
-
Syed Billah, College of Information Sciences and Technology
-
Gloria Soto, College of Health and Human Development
“Transparent and Reproducible AI-assisted Annotation for Social Data: A Multi-Agent framework”
This project will develop AI tools to make large-scale analysis of public officials’ communication more transparent and reproducible. To address slow, costly annotation and inconsistent LLM outputs, the team will create a dual-agent annotator–inspector system and a human-in-the-loop framework for faster, accurate codebook refinement.
-
Qunhua Li, Eberly College of Science
-
Yuehong "Cassandra" Tai, College of the Liberal Arts