Key points are not available for this paper at this time.
Artificial intelligence (AI) chatbots are useful in nursing education for communications training, but their effectiveness may depend on embodiment. In this pilot study, we examined whether a humanoid robot mediating an AI chatbot influences nurse–AI communication during simulated clinical history-taking. Six registered nurses completed a scenario with an AI chatbot, and a humanoid robot, whose gestures and posture were controlled via a Wizard-of-Oz method. Outcomes were task performance, perceived anthropomorphism, and speech sentiment. Participants took part in a 5-minute semi-structured interview, with key themes informing interpretation of our quantitative analyses. Bayesian regression suggests robot gestures improved task performance but reduced anthropomorphism, and AI and participant speech sentiment were positively associated. Despite our small sample size (n = 6), our findings highlight the need to carefully align robot movements with AI dialogue when designing educational tools, because incongruence between robot movement and AI speech is a barrier to learner acceptance and performance.
Taylor et al. (2025) studied this question.