Media Lab @ UIST 2026
A variety of Media Lab community members are presenting work at UIST 2026, the ACM Symposium on User Interface Software and Technology, November 2–5, 2026, at the Renaissance Center in Detroit, Michigan. The interdisciplinary conference brings together international researchers, designers, developers, and practitioners to present and discuss innovations in human-computer interfaces.
The Media Lab's contributions span a workshop, papers, posters, demos, and a keynote debate on how interactive systems should balance automation with human control.
Monday, November 2
Workshop: Bobapedia: A Workshop on Archiving Living Flavor Experiences @ 9 am ET
Presented by Professor Paul Pu Liang, with collaborators from National Taiwan University, the University of Tokyo, and MIT, Bobapedia is a participatory workshop on collaborative flavor creation with boba tea. Participants use an LLM-supported Flavor Wiki to record their sensory impressions and cultural associations, then iteratively design their own customized drinks.
Tuesday, November 3
Presented by Fluid Interfaces graduate student Valdemar Danry, Professor Pattie Maes, and collaborators from Microsoft Research, this gaze-grounded assistant uses egocentric video with gaze overlays to find where a reader is struggling. In a controlled study with 36 participants, it was rated more accurate and personalized than a text-only LLM assistant..
Poster: Murmurations: A Wearable Piezoelectric Silent Speech Interface @ 6 pm ET
Presented by Responsive Environments graduate student Sam Chin, Professor Joseph A. Paradiso, and collaborators from MIT and EPFL, Murmurations is a wearable chin sensor that captures the vibrations of silent speech using piezoelectric film. It is fitted to each user's face through 3D scanning and maps the signals to text, offering private communication for people with speech impairments.
Presented by Professor Paul Pu Liang, with collaborators from MIT, Text-to-Assembly turns natural language prompts and AI-generated 3D meshes into multi-part assemblies a robot can build. A vision-language model decides which regions become structural panels and which serve functional roles, and in a study with 32 participants, its assignments were judged functionally appropriate 90.6% of the time.
Presented by Tangible Media alum Jack Forman, Professors Hiroshi Ishii and Neil Gershenfeld, and collaborators from the Northeastern University, this demo brings the knitted shape-changing textiles and morphing garments from KnitARobo to show in person.
Wednesday, November 4
Keynote: Direct Manipulation vs. Interface Agents: The Debate Continues @ 11 am ET
Keynote debate with Professor Pattie Maes and Professor Ben Shneiderman (University of Maryland), moderated by Professor Cliff Lampe (University of Michigan). Maes and Shneiderman revisit their late-1990s arguments for interface agents and direct manipulation, and examine what today's AI-infused interfaces demand from designers, researchers, and system builders. The session invites the UIST community to reconsider how automation, personalization, and delegation should be balanced with transparency and human control.
Presented by Tangible Media alum Jack Forman, Professors Hiroshi Ishii and Neil Gershenfeld, and collaborators from the Northeastern University, KnitARobo builds on the FibeRobo project to produce shape-changing textiles with industrial knitting machines. By integrating programmable fibers into established textile manufacturing, it moves responsive, morphing fabrics closer to production at scale.
Thursday, November 5
Presented by Fluid Interfaces graduate students Yasith Samaradivakara and Valdemar Danry, and Professors Paul Pu Liang and Pattie Maes, this LLM-based approach predicts how a person is likely to speak in everyday conversations. Built on over 1,000 hours of smartwatch recordings from 14 participants, it shows that person-specific verbal behavior can be predicted from longitudinal conversational data.