Fireside · 11:00 · Convergence

A Matter of Trust

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

About this session

Daniela Rus, Director of MIT CSAIL, on AI that acts in the physical world, and what it will take for people to trust it.

On stage

moderatorAaron PressmanReporter, The Boston Globe
speakerDaniela RusDirector, MIT CSAIL
speakerJoyce SidopoulosCo-Founder & Chief of Operations, MassRobotics

Signals from this session

1:00 in the recordingDaniela Rus and her students found direct LLM-to-robot connectivity unreliable and difficult to scale. Experiments connecting robots directly to transformer-based language models showed limited reliable and scalable performance, partly because the systems lack physical-world simulation while needing to understand physics, satisfy task requirements, operate at low latency, and use energy efficiently.
1:43 in the recordingDaniela RusMIT CSAIL research found direct LLM-to-robot connectivity remains limited in reliable, scalable operation. MIT CSAIL researchers found that connecting transformer-based language models directly to robots has limited reliable and scalable capability because the models lack physical-world simulation needed for physics understanding, task requirements, low-latency operation, and energy efficiency.
4:00 in the recordingAn AI engineer system is being developed to generate robot mechanisms from high-level specifications. The AI engineer converts natural-language requirements, such as creating a robot hand optimized for a pipette or syringe, into mechanism designs with appropriate degrees of freedom.

Members see another 15 signals from this session. To read them, subscribe or sign in.

Viewpoints from this session

1:23 in the recordingDaniela RusDirectly connecting large language models to robots remains quite limited in reliability and scalability.
1:43 in the recordingDaniela RusTransformer-based statistical models lack the physical-world simulation needed for robots to understand physics and task requirements while operating with low latency and energy efficiency.
2:53 in the recordingDaniela RusLanguage was crucial to humanity's rise because it enabled cooperation, explanation, and imagination about future events.

Members see another 29 viewpoints from this session. To read them, subscribe or sign in.

Other sessions

10:30KeynoteComing HomeColin Angle, Familiar Machines & Magic10:30PanelManipulationLenore Rasmussen, Ras Labs · Alfredo Millán, Berkshire Grey · Rob Weeks, Festo · moderated by Tom Ryden, MassRobotics10:30PanelSimulation, Data and World ModelsAdam Chikha, Asynchronics · Mike Flaxman, Sand Technologies · Cecilia Liu, IntuitiveMotion · Sriram Krishnan, Istari Digital · moderated by Rachael Harkavy, Link Ventures11:00PanelReliable Autonomy in the FieldDavid Barnes, Empowering AI · Eddie Bernardon, AIPEX Racing · Christopher Metzler, University of Maryland · Pablo Berlanga Boemare, Robot Everest11:00FiresideAutonomous LabsArjun Bindu Sunil, Sahas Labs · Bingda Li, Whelix · Anna Marie Wagner, Transfyr · Chase Olle, Robot on Rails · moderated by Oded Maron, MassRobotics11:00TalkFixing American FactoriesPavan Pant, Tristar 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.
↑