2026 Winter Robotics Colloquium: Marynel Vázquez (Yale University)

Why It Matters

Understanding and engineering social context enables robots to act safely and responsibly alongside humans, unlocking broader adoption in homes, workplaces, and public spaces.

Key Takeaways

• Generalist robots require both physical and social intelligence.
• Social context defined by agents, environment, and their relationships.
• Large language models can expand robot adaptability in novel interactions.
• Experiments show robots can influence human behavior against abuse.
• Group dynamics amplify robot-mediated social influence in HRI.

Summary

In this colloquium, Marynel Vázquez of Yale University argues that the next wave of generalist robots must combine sophisticated manipulation abilities with genuine social intelligence. Using the household robot "Rosie" as a running example, she illustrates how future robots will need to interpret nuanced human cues—beyond verbal commands—to act appropriately in homes, factories, and care settings.

Vázquez proposes a concrete definition of "social context" as the set of attributes of agents, their environments, and the relationships linking them. She highlights the massive “long‑tail” of rare, ambiguous situations that current machine‑learning pipelines cannot cover, and points to large language models as a promising tool to endow robots with flexible, context‑aware reasoning. The talk also stresses that HRI research has historically fragmented the notion of context, prompting her team to unify it under this triadic framework.

A striking portion of the presentation details experimental work on robot abuse. In a study where a confederate verbally and physically mistreated a robot, participants intervened more often when the robot displayed vulnerability—shutting down briefly or expressing sadness—than when it remained passive. In a follow‑up group task, three robots collectively showed distress toward a mistreated peer, prompting a bystander to say, "Don't break it," demonstrating emergent social conformity effects.

These findings suggest that designing robots for real‑world deployment will require integrating social perception, ethical reasoning, and large‑model inference to navigate unpredictable human behavior. Emphasizing group dynamics and contextual awareness could improve safety, user acceptance, and the overall effectiveness of human‑robot collaboration across industries.

Original Description

Title: Beyond Physical Intelligence: Why Generalist Robots Require Social Intelligence

Speaker: Marynel Vázquez (Yale University)

Date: Friday, February 27 2026

Abstract: As the robotics industry moves toward deploying generalist agents in unstructured human environments, such as homes and workplaces, the research focus has largely remained on physical intelligence. While mastering physical tasks is essential, social intelligence is a critical missing piece for widespread technology adoption. To be truly effective, robots must understand, navigate, and manage the nuances of interpersonal interaction. In this talk, I will discuss two aspects of social intelligence that are fundamental for Human-Robot Interaction (HRI). The first one pertains understanding social phenomena that emerges in group interactions and that robots can potentially leverage to navigate complex social situations. The second one is implicit human feedback, i.e., communicative signals that are given off “for free” by humans and that require interpretation. Robots can leverage such implicit feedback to predict how people perceive them and better collaborate with users. Finally, I will reflect on how the latest advancements in machine learning are fundamentally reshaping the way we approach research in Human-Robot Interaction.

Bio: Marynel Vázquez is an Assistant Professor in Yale’s Computer Science Department, where she leads the Interactive Machines Group. Her research investigates fundamental problems in Human-Robot Interaction and Artificial Social Intelligence, often motivated by challenges or opportunities that arise in group human-robot interactions. Marynel received her bachelor's degree in Computer Engineering from Universidad Simón Bolívar in 2008, and obtained her M.S. and Ph.D. in Robotics from Carnegie Mellon University in 2013 and 2017, respectively. Before joining Yale, she was a collaborator of Disney Research and a Post-Doctoral Scholar at the Stanford Vision & Learning Lab. Marynel received a 2024 AFOSR YIP Award, a 2022 NSF CAREER Award, two Amazon Research Awards, and a Google Research Scholar award. Her work has been recognized with a 2025 IJCAI Early Career Spotlight, best paper awards at ACM/IEEE HRI 2023 and IEEE RO-MAN 2022 as well as nominations for paper awards at ACM/IEEE HRI 2021, IEEE IROS 2018, and IEEE RO-MAN 2016.

This video is in the process of being closed captioned.