James Landay: What’s next in human-computer interaction?

Episode Summary

Executive Summary: The episode explores how human-computer interaction is shifting from rigid interfaces to ambient, AI-enabled systems that can better support behavior change, education, and well-being. Stanford’s James Landay argues for human-centered design, emphasizing need-finding, iterative prototyping, and privacy-aware systems that augment people rather than replace them. Examples include glanceable fitness feedback, narrative-based learning tutors, and adaptive buildings.

Main Topics: Human-centered design as the foundation of HCI (Priority: 5/5): Landay argues that technology should begin with understanding user needs and real-world problems rather than starting from the capabilities of a new tool. This is framed as essential for making AI useful and trustworthy. Technology push vs. technology pull (Priority: 4/5): The discussion contrasts building from a new technology and then seeking applications versus identifying a human problem first and then selecting the right technology. In research, Landay says both approaches are often necessary. Behavior change through glanceable, ambient feedback (Priority: 5/5): Landay describes research such as UbiFit and Huizuki, where subtle visual or tactile cues on phones and wearables encourage exercise and sustainable behavior by increasing awareness and supporting long-term engagement. Narrative-based intelligent tutoring (Priority: 5/5): The conversation covers educational systems that use story as motivation for learning, including projects inspired by The Diamond Age. These systems aim to teach math and science through interactive narratives, AI chatbots, and adaptive pacing. Keeping users engaged over time (Priority: 4/5): A major challenge is abandonment of behavior-tracking devices and educational tools once novelty fades. The team experiments with narrative arcs, visual chapters, and psychologically informed design to sustain motivation. Hybrid physical-digital spaces and ambient computing (Priority: 4/5): Landay discusses how buildings and environments can be designed to influence stress, belonging, creativity, and productivity, while remaining privacy-preserving and avoiding creepy surveillance.

Key Arguments: Human-centered design should start with observed user needs, not assumptions about what technology can do. User needs are often not clearly articulated; qualitative observation and interview work are necessary to uncover the real problem. HCI is partly an art rather than a pure science because need-finding relies on interpretation, not just statistics. Behavior change works better when feedback is subtle, glanceable, and integrated into existing routines rather than demanding attention. Long-term engagement requires more than novelty; narrative structure and positive psychology can help prevent abandonment. AI tutors should augment teachers and learners, not replace them, by adapting content and offering help when students get stuck. Built environments can be designed to improve well-being, but only if systems remain privacy-preserving and non-creepy.

Data Points: Phone checks per day: 100–200 times - Used to justify glanceable feedback opportunities in behavior-change systems. Behavior-tracking abandonment after 3 months: About one-third quit - Landay cites Fitbit-style devices losing users after the novelty wears off. Behavior-tracking abandonment after 6 months: 50% abandoned - Illustrates the challenge of long-term engagement with wearable trackers. Time spent in built environment: 87% of the day - Cited as motivation for studying hybrid physical-digital spaces. Narrative structure in Huizuki: 13 visual chapters - The story-based behavior-change interface uses a multi-chapter narrative arc to sustain attention.

Pivotal Quotes: "The key to doing this in a thoughtful way is to put people first." — James Landay: Landay summarizes the core principle of human-centered design. "It's not a science. It's more of an art." — James Landay: He explains why need-finding and user observation are difficult to formalize. "Don't make it such that people can't move on if they can't master something. Instead, make it about exposure." — Dan Schwartz (quoted by James Landay): Advice shaping the design of educational narratives and learning activities.

Implications: The future of AI interfaces depends on human-centered, privacy-aware design that supports behavior, learning, and well-being without overwhelming users. Successful systems will be subtle, adaptive, and engaging over time.