Teacher-Facing AI in Education: A Systematic Review of Impact and A Taxonomy of Teacher-AI Teaming

Abstract

Artificial intelligence (AI) is increasingly integrated into educational practice, promising to enhance teaching and learning. Yet delegating pedagogical tasks to AI raises concerns about teacher deskilling, erosion of professional judgement, and diminished agency. Drawing on the teacher-AI teaming taxonomy, we conducted a systematic review of 103 studies of teacher-facing AI tools across educational contexts to analyse system capabilities, patterns of interactions and influences on teaching practice. Our analysis reveals that advanced technical capabilities of contemporary AI models were utilised at lower transactional teaming levels for automating narrow instructional functions. Situational and operational teaming, which supports teacher awareness and teacher-directed goal execution, are also common in Generative AI in education literature and have demonstrated complementary benefits. However, available evidence relies on student-focused measures, whereas the augmented impacts, particularly on teaching effectiveness, remain underexplored. In contrast, higher forms of teaming as praxical and synergistic teaming, which afford co-adaptation and structured co-reasoning, promoting reflective professional practice, remain rare. These findings suggest that current teacher-facing AI systems have yet to fully leverage AI to strengthen teacher agency. We conclude the paper with recommendations on how to realise such forms of teaming, which require advances not only in model capability but also in interaction design that support transparent reasoning, teacher-controllable interfaces, and sustained teacher participation in system development.