What the paper studied
This research investigated how three anthropomorphic features of restaurant service robots—voice pitch (high vs. low), degree of humanlike appearance (high vs. low), and perceived gender (male vs. female)—influence customer perceptions of robot attractiveness. Grounded in Interpersonal Attraction Theory, the authors measured attractiveness across three dimensions: physical, task, and social attraction, and tested how each shaped customers' willingness to interact with the robot. Two scenario-based online experiments were conducted with US participants recruited through Prolific, using AI-generated robot imagery paired with AI text-to-speech voice-over monologues in a casual-dining welcome encounter. Study 1 (N=463) crossed voice pitch with humanlike appearance; Study 2 (N=463) crossed voice pitch with perceived gender.
Key findings
- Voice pitch was the only manipulated feature to produce a significant multivariate effect on attractiveness across both studies.
- Low-pitched voices were rated significantly higher than high-pitched voices on task attraction and social attraction in Study 1, and on physical attraction and social attraction in Study 2.
- Degree of humanlike appearance had no significant effect on any dimension of attractiveness, challenging the assumption that visual anthropomorphism drives acceptance.
- Perceived gender (male vs. female) also had no significant effect on attractiveness, and did not interact with voice pitch.
- Physical, task, and social attraction all significantly predicted willingness to interact with the robot (p < 0.001).
Why it matters for hospitality
The results reframe how operators should think about service robot design in dining settings. Rather than investing in expensive, highly humanlike robot exteriors or gendered avatars, restaurants stand to gain more from tuning auditory cues—particularly voice pitch—that convey competence, warmth, and trustworthiness. Because restaurant service encounters tend to be brief and transactional, guests appear to prioritize functional cues over visual representation, and gender stereotypes carry less weight than in longer or more personalized service roles. This shift has real cost implications: non-humanoid robots with well-tuned voices may deliver comparable guest acceptance at a fraction of the price of high-fidelity humanoids, while also reducing staff anxiety about being replaced by lifelike substitutes.
Practical takeaways
- Prioritize low-pitched, humanlike voices when programming or selecting restaurant service robots to increase perceived competence, warmth, and willingness to interact.
- Do not overspend on highly humanlike physical designs for casual-dining roles—guests do not reward the added realism with more positive perceptions.
- Treat robot gender as a low-stakes design decision in restaurants; gender-neutral robots are unlikely to hurt attractiveness compared to explicitly male or female designs.
- Position robots as functional helpers to staff (routine or repetitive tasks) rather than humanoid replacements, which is both more cost-effective and easier for employees to accept.
- Findings apply most directly to casual-dining welcome encounters; fine dining, quick service, and longer-form hospitality contexts (hotels, healthcare) may weigh visual and gender cues differently and warrant separate testing.