The objective buffer: comparing embodied artificial intelligence agents and human evaluation acceptance of negative performance feedback

The integration of embodied artificial intelligence agents (EAAs), such as social robots, presents a novel opportunity for scaling personalized education. However, the effective deployment of EAAs for critical tasks, like delivering negative performance feedback, hinges entirely on user acceptance. Drawing on the Computers Are Social Actors paradigm and the technology acceptance model, this study investigates how the evaluation source (EAAs vs. human instructor) influences student acceptance of harsh, standardized critique in a public speaking context (N = 55). Results show that participants rated the social robot significantly higher than the human evaluator on credibility and social presence. Crucially, the robot’s feedback was rated significantly more favorably across almost all quality metrics (helpfulness, value, and importance) than the humans. We propose and validate the concept of the “objective buffer,” arguing that the robot’s expected neutrality minimizes the ego threat typically caused by human evaluators, thereby enhancing the cognitive acceptance and perceived usefulness of the critique. This finding demonstrates a unique pedagogical advantage for EAAs, suggesting that their optimal role is to manage emotionally sensitive, high-stakes feedback. This research offers valuable, generalizable guidance for designing EAAs and technology-enhanced learning systems in contexts requiring objective evaluation.