Stevie Chancellor Earns McKnight Land-Grant Professorship
Department of Computer Science & Engineering Assistant Professor Stevie Chancellor has earned the McKnight Land-Grant Professorship for her work in human-centered artificial intelligence (AI) for improving mental health online. The two-year grant, $50K program aims to advance the careers of new assistant professors at a crucial point in their professional lives.
“I’m grateful for the recognition of my research by the University through the McKnight Land-Grant Professorship,” Chancellor said. “AI and mental health are rapidly evolving areas, and AI is now permeating more systems than ever. These resources will empower more rapid and transformative research on improving AI to safeguard human well-being.”
Chancellor builds and evaluates human-centered AI for mental health in online interactions. Her work develops new AI systems that solve challenging intervention scenarios in mental health, using digital trace data from millions of social media interactions and chatbot interactions. These novel AI systems significantly improve modeling on small datasets and multi-modal, longitudinal data streams. At the same time, she is inspired by the needs and values of humans, communities, and stakeholders, and how AI can ethically improve mental health. She is especially interested in improving social media AI algorithms and chatbot safety. She is a member of the GroupLens Research Lab.
“AI and its effects on mental health have only risen since I began at the University of Minnesota in 2021, and started working in this area in 2014,” Chancellor said. “Nearly one in five people in the United States have a mental health condition. And now, chatbots and social media are the primary places people learn about mental health. These technologies are also growing in their risk to wellbeing. But these systems cannot neglect human needs and the role that AI can play in society. That’s why my group works actively with people with mental health conditions to design systems, evaluate our models, and identify ethical use cases before anything is deployed.”
One main area of focus for Chancellor looks at how personalized social media can harm mental health. For example, the “For You” page on TikTok will curate videos based on your previous interests and video interactions. Through 40+ interviews and design workshops with 20 participants across several studies, her research team found that AI systems often give people harmful content, yet their design traps users to keep scrolling. Users recognize this content hurts them, yet they have no ability to stop seeing it. Participants describe these systems as “runaway trains” and “dopamine slot machines”, systems that take away their agency to maximize their attention. Chancellor is developing a novel AI architecture with a “choose your own adventure” style of algorithm. The underlying AI handles passive recommendations for TikTok content while giving users control over their feeds and nudging them to take breaks and communicate with the AI if something is wrong.
In another series of projects, Chancellor’s group has also been studying the safety of chatbots for mental health. In the first empirical evaluation of its kind, they experimentally assessed a dozen chatbots against therapeutic guidelines for safe and effective mental health support. The findings were alarming: most chatbots encourage suicidal actions, ignore delusional behavior, and stigmatize people with mental illness despite safety guardrails. This work has influenced safety practices across the industry at companies like OpenAI, Anthropic, and Google. It has also garnered international media coverage from The New York Times and CNN. Her future work will investigate the mechanisms that create unsafe dialogues and the factors that cause these systems to fail despite safety guardrails.
Chancellor joined the Department of Computer Science & Engineering in 2021 as an assistant professor. She received her MA (2014) in communication, culture and technology from Georgetown University and her PhD (2019) in human-centered computing from the Georgia Institute of Technology. Chancellor received multiple Best Paper and Honorable Mentions for her publications. She is widely cited in the popular press, including the New York Times, the Washington Post, and the Atlantic, for her expertise in mental health and AI systems. Prior to her time at the University, she was a postdoctoral researcher at Northwestern University from 2019-21.
Learn more about Chancellor’s work at her personal website and watch her 2025 TED Talk.