K-12 teachers' perspectives on AI use in education through the lens of activity theory

Synthesis: Lee & Anson (2027) use activity theory to explore how Australian K-12 teachers perceive AI and its impact on education, drawing on qualitative data collected in April/May 2023 — roughly six months into the sudden global recognition of generative AI. Thematic analysis identified six major themes capturing in-service teachers' views: that AI is unsuitable, impersonal, imperfect, uncertain, assisting, and inevitable. The study theorizes activity theory in an AI world, highlighting the complex, nuanced, and often contradictory sentiments teachers hold, and positioning AI as a tool-mediated, goal-directed mediator within the school activity system.

Key Findings

• Six themes of teacher perception. Teachers characterized AI as unsuitable (for their context), impersonal (lacking relational warmth), imperfect (error-prone/unreliable), uncertain (opaque/unpredictable), assisting (useful for routine tasks), and inevitable (a force to be reckoned with).
• Activity theory as a framing. The study conceptualizes AI as a tool/mediator within the school activity system (subject, tools, object, rules, community, division of labor), arguing that digital tools — AI included — do not force action but mediate goal-directed activity; humans (not tools) retain agency.
• Contradictory, nuanced sentiments. Teachers' views were simultaneously favorable (assisting) and skeptical (unsuitable, impersonal, imperfect, uncertain), and resigned (inevitable) — a complexity that individual-belief models may flatten.
• Context sensitivity. AI as a mediating tool interacts with the specific learning environment (tools, activities, community), so its adoption and meaning vary by school context.

What this means for practice

• Instructors. Audit your own digital competency and preferred pedagogical approach before adopting AI, and map where it adds value across students' self-study, teacher-led learning, peer learning, and learning with AI.
• Instructors. Define the division of labor with AI for your own work and for students, and judge which domains in your subject benefit most from innovation and which are most vulnerable to technological threats (e.g., mathematics, reading, critical thinking).
• Instructors. Build your own AI literacy so you can recognize when AI output is unsuitable, impersonal, imperfect, or uncertain, and model that judgment for students.
• Instructors. When AI adoption stalls, treat it as a property of the whole school activity system: examine the tools, rules, community, and division of labor rather than pressing harder on training individual teachers.
• Instructors. Use the six-theme vocabulary (unsuitable, impersonal, imperfect, uncertain, assisting, inevitable) to move AI conversations with colleagues and students past a simple for/against binary.

Limitations

• Australian in-service teachers only, so the data may not reflect teacher sentiments in other parts of the world.
• Convenience sampling and voluntary participation: 66 teachers supplied qualitative responses out of 206 who completed the whole survey, recruited mainly through the authors' institutions, so they do not represent the broader population of Australian teachers and the responses skew toward certain viewpoints; the sample was also younger than the national workforce (mean age 36 versus 46).
• The data capture one moment in AI development — collected April 5 to May 22, 2023, roughly six months after the sudden public interest in generative AI.
• Only three broad open-ended questions were used, with a missing-response rate of about 15.7%; more structured questions tied to activity-theory constructs would probe deeper.

Citation

Lee, J., & Anson, D. W. J. (2027). K-12 teachers' perspectives on AI use in education through the lens of activity theory. Research and Practice in Technology Enhanced Learning, 22, 5.