Can care be simulated? Teachers’ judgments of educational messages in the age of generative AI

This exploratory mixed-methods study investigates how perceived authorship relates to judgments of educational care in messages produced by teachers and generative AI. An online questionnaire involving 807 participants in teacher qualification programmes combined evaluations of three educational vignettes with open-ended responses. Messages generally received positive assessments, although ratings declined in situations involving greater emotional vulnerability. Evaluations were consistently higher when participants attributed a message to a human teacher rather than AI. The human-authored message was also the most frequently misidentified as AI-generated. Qualitative findings associated care with contextual understanding, authenticity, human presence, and moral responsibility. Participants recognized AI’s usefulness for drafting and initial support while identifying limits to delegating relational and ethically sensitive decisions. Despite methodological limitations, the findings suggest that teacher education should address both the practical uses of AI and the human responsibility underpinning educational care.