Summary: Robots that display vulnerability change how people talk and interact within teams.

Source: Yale

Robots Showing Vulnerability Boost Team Conversation and Positive Group Experience

In a Yale-led experiment, teams of three people and a robot played a collaborative game in which the robot occasionally made errors and reacted in different ways. When the robot acknowledged mistakes, told a joke, or shared a brief personal remark—signals of vulnerability—human teammates spoke with one another more, distributed conversation more evenly, and reported a more positive group experience than teams whose robot remained silent or offered only neutral, task-related comments.

The study, published in the Proceedings of the National Academy of Sciences, explored how the behavior of social robots affects not just human–robot exchanges but also human-to-human interaction. The research team found that modest, human-like expressions of fallibility from a robot can change conversational dynamics among people working together.

Study Design and Key Findings

Researchers arranged 153 participants into 51 teams composed of three human players and one robot. Each team worked through 30 rounds of a tablet-based game that required collaboration to build efficient railroad routes. At the end of each round the robot behaved according to one of three experimental conditions: remaining silent, making a neutral, task-focused statement (for example, stating the score or the round number), or making a vulnerable remark such as acknowledging an error, sharing a short personal story, or telling a light joke. The robot occasionally lost rounds in all conditions, so its vulnerability was not a result of only winning or losing.

Teams that included a robot expressing vulnerability spent roughly twice as much time talking to one another during the game compared with the other conditions. Conversation among team members increased more when robots expressed vulnerability than when they made neutral comments. The researchers also observed a more equitable distribution of speech among teammates when the robot spoke—either vulnerably or neutrally—than when the robot remained silent.

Participants paired with vulnerable robots consistently reported enjoying the team experience more than participants in the silent or neutral-robot conditions. These subjective reports, together with the objective measures of conversational time and balance, indicate that robot speech style can meaningfully influence group dynamics and interpersonal engagement.

Implications for Design and Deployment of Social Robots

The results carry practical implications for how designers and organizations deploy robots in everyday environments. As social robots appear more frequently in public spaces—stores, hospitals, factories, and offices—it becomes important to anticipate how robot behavior will affect human teams that work alongside them.

Nicholas A. Christakis, Sterling Professor of Social and Natural Science, emphasized the broader societal question raised by these findings: as artificial agents become part of our social systems, we must consider how to design their behaviors so they strengthen rather than degrade human relationships. The study suggests that programming robots to show modest vulnerability can foster communication and a more positive team climate.

Co-author Sarah Strohkorb Sebo, a Ph.D. candidate in computer science, noted that even robots not explicitly designed for social interaction can shape workplace dynamics. For example, a distribution robot on an assembly line that consistently hands parts to a single worker could unintentionally create social friction. Insights from this study can guide the creation of robotic behaviors that encourage balanced participation and reduce social tension among human teammates.

Funding and Authorship

The research received support from the Robert Wood Johnson Foundation and the National Science Foundation. The study’s lead author is Margaret L. Traeger, a Ph.D. candidate in sociology at the Yale Institute for Network Science. Other co-authors include Brian Scassellati, professor of computer science, cognitive science, and mechanical engineering at Yale; Malte Jung, assistant professor in information science at Cornell; Sarah Strohkorb Sebo, Ph.D. candidate in computer science; and Nicholas A. Christakis.

Source:
Yale

Media Contacts:
Bess Connolly – Yale

Original Research (open access):
“Vulnerable robots positively shape human conversational dynamics in a human–robot team.” Margaret L. Traeger, Sarah Strohkorb Sebo, Malte Jung, Brian Scassellati, and Nicholas A. Christakis. Published in Proceedings of the National Academy of Sciences (PNAS).

Abstract (Summary)

The study investigates how a social robot’s spoken behavior influences communication among humans in teams. Across 51 groups of three humans plus one robot playing a collaborative game, teams with robots that made vulnerable statements conversed more with each other, distributed their talking time more evenly, and evaluated their group experience more positively than teams with a neutral-speaking robot or a silent robot. These findings indicate that robot speech patterns can shape human-to-human interactions and suggest opportunities to design artificial agents that promote more effective and inclusive group communication.