A new study found people become more suspicious of expressive humanoid robots that make errors.
When an animated robot made mistakes, participants' oxytocin levels rose, but this tracked with suspicion.
This challenges the assumption that lifelike robots earn more trust.
What happened
In a study published in Science Robotics, 50 people held conversations and made joint decisions with the commercial humanoid robot Pepper. When the robot was animated and made conversational mistakes, participants' oxytocin levels rose, but this was linked to suspicion, not affection, and they trusted the robot less and took its advice less often.
Why it matters
The findings challenge the common design assumption that lifelike, socially expressive robots earn more trust. Expressive cues appear to shift how people perceive a mistake from a technical malfunction to a social violation, engaging the brain machinery used to judge people.
What to watch
The study involved only young men and one robot design. Researchers want to test whether the same oxytocin-linked vigilance appears in women, mixed groups, other cultures, and other robot designs, and whether robots can repair trust by acknowledging errors or apologizing.
Ask the AI about this article →
The study's findings highlight a critical nuance in human-robot interaction. While the popular understanding of oxytocin links it to bonding and affection, the researchers observed that in the context of an expressive robot's errors, higher oxytocin levels were associated with decreased trust and reduced influence. This suggests the hormone was tracking suspicion, not affection, during the social violation of norms.
This research is part of a broader trend in studying trust as a multilevel phenomenon, extending beyond individual attitudes to relationships and societal networks. The use of wearable brain imaging systems like functional near-infrared spectroscopy allows researchers to study social cognition in natural encounters, which is not possible with conventional MRI scanners.
The study also opens avenues for future research, including testing whether the observed vigilance appears in diverse populations and whether robots can repair trust by acknowledging errors or apologizing, similar to human interactions.
For example, today's edition would include:
AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.
Free · takes 30 seconds · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. Q&As are published on this page for other readers too.
A UK study by UK AI Security Institute and Limbic AI surveyed 6,474 British adults
Broadcom's Clayton Donley says companies are doing mission-critical work with AI agents quickly, but without t…
Bank of England governor Andrew Bailey warned that advanced AI poses risks to financial infrastructure in a le…
As AI agents perform real business tasks, 'Agentic Identity' (giving each AI a unique employee-like ID) and 'D…
Andrew Bailey, head of the world's financial stability watchdog, warned in a letter to G20 finance ministers a…
Skild AI has introduced S1, its flagship robot foundation model, which lets robots learn new, complex tasks by…
