‘AI does not get tired, but it has no feelings’: feedback quality, relational motivation, and learner agency in Thai EFL academic writing

Abstract

As generative AI tools become increasingly embedded in second language writing instruction, understanding how learners experience and reconcile AI-generated feedback alongside traditional teacher feedback has become an urgent pedagogical concern. This study investigated how Thai EFL university learners perceive, emotionally engage with, and negotiate AI-generated and human teacher feedback in academic writing. Inductive thematic analysis of semi-structured interviews and focus group discussions with 26 undergraduate students revealed three primary findings. First, the preference for AI feedback reflected its systematic coverage and the institutional constraints governing teacher commentary. This suggests that the structural limitations of human instruction under institutional constraints, rather than inherent evaluative superiority, may account for participants’ perceived advantages of AI. Second, participants described teacher feedback as motivationally irreplaceable due to its relational dimensions. Conversely, AI appeared to serve an unanticipated threshold utility. This role reduced the emotional cost of early help-seeking and facilitated focused subsequent engagement with instructors. Third, cultural norms of deference in Thai educational contexts appeared to act as a barrier to learner agency. These norms appeared to restrict students’ capacity for action even when participants regarded AI feedback as more credible than teacher grades. These findings suggest a boundary condition for distributed agency theory in high-power-distance settings, challenge foundational assumptions in hybrid feedback research, and reposition the affective role of AI within the feedback ecology. Implications are drawn for feedback theory, writing pedagogy, and AI tool design.

The authors' abstract, as published at the source. Innovation in Language Learning and Teaching, 2026 · DOI ↗