Accessibility Lunch - Riku Arakawa
October 5, 2026 12:30PM—1:30PM
Location:
In Person and Virtual - ET
-
Newell-Simon 3305 and Zoom
Speaker:
RIKU ARAKAWA,
Ph.D. Student, Human-Computer Interaction Institute, Carnegie Mellon University
https://rikky0611.github.io/
Context-aware proactive systems can move beyond today’s reactive, chatbot-style interactions by sensing users’ activities and states and providing support at appropriate moments. Such capabilities are particularly promising for health and well-being, where AI assistants could serve as companions that understand users’ conditions and evolving needs in everyday life. However, real-world sensing is inherently uncertain: sensor data are noisy, human behavior is ambiguous, and the meaning of an observed activity varies across individuals and contexts. As a result, simple sensing-and-intervention pipelines can trigger inappropriate interventions, frustrate users, and undermine trust. My research investigates how to build proactive systems that remain reliable under sensing uncertainty through three complementary directions: (1) understanding and mitigating sensing uncertainty, (2) designing uncertainty-aware Human-AI interactions, and (3) enabling adaptation through everyday interactions. In this talk, I will illustrate this research trajectory through several interactive assistants for health and well-being. Drawing on deployments involving more than 300 patients, I will share lessons from translating these systems into real-world settings and discuss future directions toward reliable, adaptive AI companions for everyday care.
For More Information:
pmowar@andrew.cmu.edu