A lot of research on AI focuses on what it helps people produce. I'm interested in how interaction with AI changes the way people understand the problem itself. This is an earlier concern.
Across creative work, decision-making, and everyday problem-solving, people rarely begin with a fully formed objective. Goals evolve. Constraints shift. New possibilities emerge. Through repeated interaction, AI can become part of these processes.
These shifts shape what gets created, what gets pursued, and ultimately what gets solved.
Before influencing solutions, interaction with AI influences goals, constraints, assumptions, and problem definitions. A person might begin a task with one understanding of a problem and gradually arrive at another. Sometimes these shifts are productive. Sometimes they are subtle enough that people don't recognize they are occurring. Understanding how and when these changes happen is central to my work.
Actual use of AI often extends beyond isolated prompts and responses to a more iterative interaction. Through extended conversations, ideas are revisited, refined, abandoned, and rediscovered. Small suggestions can accumulate into meaningful changes in direction. I'm interested in how understanding develops across these interactions and how AI influences the trajectory of thought over time.
Important outcomes can emerge from the interactions between people and AI differently than either participant independently. This perspective raises different questions about evaluation. Beyond accuracy or output quality, I'm interested in understanding how AI affects exploration, framing, judgment, confidence, and creative autonomy. These effects may be especially important in domains where goals are uncertain and the work helps define what success looks like.
As AI becomes embedded in consequential decisions, understanding its effects on reasoning becomes increasingly important. AI can help people identify possibilities they might miss otherwise. It can also shape attention in ways that shape what information is considered relevant, what alternatives are explored, and what conclusions feel plausible. I'm interested in finding out how these influences occur, how people recognize them, and how we can design systems to support human judgment without quietly replacing it.
My current research examines how people and AI construct problems together during creative tasks. Through a series of studies, I investigate how individual differences, interaction patterns, and AI behavior influence the evolution of goals, frames, and ideas over time. A bit more broadly, I'm interested in the design of human-AI systems that support exploration, reflection, and thoughtful decision-making while preserving human agency.