Older adults’ perceptions of robots with differing intelligence in social multi-robot interactions

Social robots present significant potential to support ageing societies, yet their integration into older adults’ lives still requires substantial investigation. Existing research has explored multiple factors that influence older adults’ attitudes towards robots, but few have directly manipulated perceptions of robot intelligence. We address this gap on the premise that older adults may experience reduced confidence in managing technology, making them more sensitive to cues of robot competence, as the basis for their trust and acceptance. This study adopts a human-multi-robot approach, situating social interactions within a group where human-robot and robot-robot interactions co-exist. Unlike research limited to dyadic interactions or that deploys multiple robots in isolation to provide complementary assistance or different points of engagement, we model group (triadic) interactions in a structured experiment where participants played a cognitive game simultaneously with two social robots of different cognitive competence. This design was intended to leverage inter-robot comparisons when forming older adults’ perceptions, rendering the manipulation of perceived intelligence more salient. Using a mixed-method approach, we explored how the intelligence construct affected trust, usability, and the willingness to adopt social robots, including preference for continued interaction. Our findings showed participants placed greater trust and intent to use the robot with higher constructed intelligence, though many favoured the robot that appealed to them aesthetically or displayed human-like fallibility, which some perceived as intelligence. The interaction increased participants' willingness to use robots in other proxy (sensitive, health-related) contexts, suggesting that perceptions formed in one context may transfer to others, potentially reinforcing acceptance of robots broadly with continued interactions.