Realizing AI for

REALIZE Lab is a research group at the University of Michigan, led by Elizabeth Bondi-Kelly.

We develop use-inspired AI through advances in machine learning and multi-agent systems. We study how people and AI can work together, involving communities in design and experts in guiding and evaluating models. Our methods account for limited, noisy data and incorporate physical knowledge.

Eight REALIZE Lab members smiling together on a footbridge
Meet the team →

Research areas

Health & Healthcare

We build and evaluate AI for patients and clinicians, from prenatal-care agents and reproductive-health information to cancer-note comprehension and medical imaging. We study misinformation, trust, and when people rely on AI advice.

View publications →

Environmental Health & Conservation

We develop machine learning for environmental monitoring and wildlife conservation: mapping PFAS contamination, guiding geospatial sampling with limited data, and enabling experts to improve image retrieval for wildlife re-identification.

View publications →

Human-Centered & Participatory AI

We study how people shape AI systems, from recruiting community participants to translating stakeholder needs into system requirements. Our work develops approaches to human-driven AI design and evaluates how people understand and use AI outputs.

View publications →

Physics-Informed Machine Learning

We incorporate mechanistic knowledge into machine learning for physical systems. Our work on stochastic transport learns from both observed outcomes and how systems respond to changes in physical parameters, supporting prediction under interventions.

View publications →

Selected work

All 14 publications →