Can Machines Perform Emotional Labor in Education? Tracing Debates in Early Philosophy of Artificial Intelligence

Can an artificial intelligence (AI) system perform the emotional labor of teaching, or can it only reproduce emotional labor's outward signs? In this article, I argue that the latter does not by itself justify the former. Emotional labor in education includes surface acting, but it still presupposes an agent capable of emotion, self‐regulation, and responsibility within a pedagogical relation. Its educational value can also depend on students' reasonable perception that care is not wholly fabricated. I develop this claim through selected debates about machinic emotion from 1950 to 1996. Ziff, Smart, and Putnam dispute whether feeling requires biological life or can be attributed through organized behavior; Weiss, Scriven, and Danto test whether relation, honesty, and language can evidence machine emotion; and Dyer, Ortony et al., and Arbib contest whether computational appraisal can capture embodied affect. These debates do not supply a historical test that contemporary systems must pass. They expose three still‐relevant justificatory questions about ontology, evidence, and embodiment that divide educational critics and technologists. This article uses those questions to distinguish computational representations and simulations of emotion from warranted claims that a system has emotion and performs emotional labor. The result is a shared vocabulary for evaluating affective AI in education and for identifying which forms of pedagogical care cannot simply be inferred from emotionally convincing output.