Fireside · 15:30 · Models Scale

Out of the Lab

Minds in Motion at MIT, Physical AI Summit · 30 Oct 2026 · Cambridge, United States

About this session

A short talk and a conversation on how machines learn to predict and act in the physical world, and what it takes for that learning to hold up in the real world.

On stage

moderatorJohn WernerFounder & CEO, Imagination in Action
panelistPulkit AgrawalAssociate Professor at MIT, Eka Robotics
panelistNa LiProfessor of Electrical Engineering & Applied Mathematics, Harvard University

Signals from this session

5:09 in the recordingHarvard University is developing a mixed-reality AI assistant for laboratory scientists. The system uses Mixed Reality Google Glass to let the AI observe what scientists see and do, while scientists can ask it to record procedures, perform analyses, and investigate anomalies during experiments instead of waiting until the experiment ends.
5:41 in the recordingA model trained on internet-scale data outperformed many specialized robotics foundation models. The study found that the internet-trained model performed better than many foundation models developed specifically for robotics, indicating broader generalization from large-scale internet data than from specialized robotics training alone.
6:02 in the recordingHarvard University demonstrated Cinex providing real-time laboratory guidance through Google Glass. In a clean-lab fabrication workflow, a scientist wearing Google Glass received AI-generated knowledge and real-time analysis overlaid on the view, including recognition of hazardous chemical combinations and an immediate warning not to combine them.

Members see another 38 signals from this session. To read them, subscribe or sign in.

Other sessions

15:00PanelSide by SideDavid Dixon, MIT Open Learning · Aleksandra Przegalinska, Kozminski University · Peter Danenberg, Google DeepMind · Alp Uguray, OpenAI · moderated by Jon Chesto, The Boston Globe15:00PanelFrontiers of Physical AICamron Blackburn, MIT · Erik Strand, MIT Center for Bits & Atoms · Jake Read, MIT · moderated by Neil Gershenfeld, MIT Center for Bits & Atoms15:00PanelEnterprise AI Meets the Physical WorldSatya Nitta, Emergence AI · Thiru Venkatachalam, IBM Consulting · Roman Nemet, The Digital Loom · moderated by Gabriela Torres V., Massachusetts AI Hub15:30PanelSea, Air and StreetJill Wittels, Nexteon Technologies · Sanjiv Singh, Near Earth Autonomy · Greg Lynn, Piaggio Fast Forward · Kade Crockford, ACLU of Massachusetts · moderated by Tucker Hamilton, 96th Ops Group15:30PanelRobots That BuildVikas Enti, Reframe Systems · Mauricio Pereira, Industrial Automata · Michael Santora, Meta Logic · moderated by Frazer Anderson, Link Ventures15:30TalkTurning Humanoids into Useful Robot WorkersMichael Chen, Yondu AI

Full program →

Members read every signal and every company file.

They also get the Frontier Tech Digest every morning and can ask their own questions. What Frontier Tech Intelligence is →

Subscribe · $99 a month
Independent coverage, not affiliated with any organizer.
↑