Safe and Compassionate ML Recommendations for People with Mental Illnesses
PI: Stevie Chancellor
Assistant Professor of Computer Science & Engineering
University of Minnesota
Framework component: Interaction
Content recommendation drives much of the engagement, connection, and discovery of new information on modern social platforms. However, these recommendations can be a double-edged sword for people with mental illnesses. This project’s research goals are to: 1) Identify sociotechnical factors that make recommendations helpful or harmful to people with mental illnesses on digital platforms and possible solutions; and 2) Design and evaluate one participant-centered ML intervention to alleviate these harms.
The project will conduct participatory workshops with participants who have or have had psychosocial mental illnesses (e.g., depression, anxiety, suicidal ideation, eating disorders) and have experienced algorithmic harm on social networks such as TikTok. Design workshops will have speculative and participatory exercises on what better recommendations could look like. The project will build a system that makes safer and more compassionate recommendations for people in distress and deploy it with a small experiment of 8-12 participants.
Key Personnel
Fernando Maestre
Professor of Computer Science & Engineering
University of Minnesota
Ashlee Milton
Graduate Student, Computer Science
University of Minnesota
Loren Terveen
Professor of Computer Science & Engineering
University of Minnesota