We forecast at two scales: population-level effects to guide study planning, and individual-level agent models to anticipate behavior and map heterogeneity.
We test LLM-generated messages and structured dialogues, evaluate transparency and disclosure, and propose guardrails.
We study AI’s role in psychological well-being through brief, structured interactions grounded in behavioral science.
We develop a public interest roadmap and pilot generative AI to improve social service delivery, from resource navigation to SNAP and TANF processing.
