The Adaptive Behavior Systems Department builds human-AI systems that understand people and change real-world behavior. We combine behavioral science, causal reasoning, generative models, and large-scale field studies to (1) identify behavioral insights, (2) build models of human behavior, and (3) design and safely deploy personalized, scalable interventions—from EV-charging studies to adaptive in-cabin interfaces. By creating tools that generate and pretest interventions, benchmarking causal AI, and prioritizing safety and trust, the department accelerates the adoption of carbon-neutral technologies, improves driver and passenger safety, and makes AI assistance useful and responsible at scale.

Our team takes a human-centered approach to developing AI systems encouraging behavioral changes.

Our solutions provide capabilities to understand, predict, and influence human behavior—whether that means accelerating the adoption of carbon-neutral technologies or designing safe, personalized interactions between people and AI agents.

Our customers are designers, researchers, engineers, and decision-makers who need faster, more reproducible human insight and a reliable path from idea to impact. We help teams scale behavioral solutions, reduce deployment risk, and accelerate the adoption of carbon-neutral technologies and safer vehicle experiences.

In the Adaptive Behavior Systems Department, we are focused on three core challenges:

How can we move beyond correlation and build AI systems that truly understand cause and effect in human behavior?
Real-world decisions depend on predicting consequences—not just patterns. We must develop models that can simulate interventions, anticipate unintended effects, and reason about “what if” scenarios with rigor and reliability.

How can we design behavior-change interventions that are personalized, effective, and durable at population scale?
Small pilots are not enough. We need systems that identify who responds to what, under which conditions, and for how long—so interventions can scale across diverse users, markets, and contexts.

How can we deploy AI-driven behavior change safely, responsibly, and with trust?
As AI systems become more proactive and adaptive, we must ensure transparency, fairness, user agency, and measurable real-world benefit—especially in safety-critical environments like vehicles.

These challenges sit at the intersection of behavioral science and advanced AI. Generative models allow us to rapidly create and test candidate interventions. Causal modeling enables simulation of outcomes before deployment. Large-scale field studies provide the evidence needed to validate durability and equity. The central question is not just whether AI can change behavior—but whether it can do so reliably, ethically, and at scale.

Image

Representative Publications

• Promoting Sustainable Charging Through User Interface Interventions
• Save A Tree or 6 kg of CO2? Understanding Effective Carbon Footprint Interventions for Eco-Friendly Vehicular Choices
• Using LLMs to Model the Beliefs and Preferences of Targeted Human Populations
• Causal AI Framework for Unit Selection in Optimizing Electric Vehicle Procurement

Press

• UW Accessibility Research Could Help Make Cars Safer