Study finds 87% of teachers abandon AI agents due to system design, not skill gaps
Systemic contradictions, not lack of know-how, drive teacher disengagement from AI tools.
A new study from researchers at multiple institutions, including Haiyang Xin and colleagues, tackles the paradox of why teachers disengage from creating pedagogical AI agents after completing professional development. In Cycle 1 (N=218), behavioral tracking and interviews showed that 87% of teachers ceased agent creation within three weeks—despite finishing comprehensive training. The root cause was not a lack of technical skills but systemic contradictions in their work environment that frustrated basic psychological needs.
Cycle 2 (N=26) applied a combined Cultural-Historical Activity Theory (CHAT) and Self-Determination Theory (SDT) framework to redesign the professional development program. The intervention directly targeted the diagnosed contradictions and achieved synchronized gains in both teacher capacity and willingness to build AI agents. The authors argue that implementation failure is often a rational response to need-thwarting systems, and they offer a replicable CHAT-SDT diagnostic framework for transformative professional development in educational AI.
- 87% of teachers stopped creating AI agents within 3 weeks after professional development, despite completing the training.
- Cycle 1 analysis of 218 teachers found systemic contradictions, not skill deficits, were the primary cause of disengagement.
- Cycle 2 redesign using CHAT and SDT frameworks for 26 teachers improved both competence and willingness to build AI agents.
Why It Matters
Edtech leaders must redesign training environments, not just content, to sustain teacher AI adoption.