Pip Sanderson -'Developing a Position on AI in TeacherEd', GenAI in TeacherEd Summit 2025

In this session, Pip Sanderson (National Institute of Teaching) offers a deeply reflective and intellectually grounded exploration of what it means for trainee teachers to use AI with thoughtful agency. Pip begins by acknowledging how emotionally and professionally turbulent the AI landscape can feel, introducing the idea of being “ironsick” – a word describing the disorientation that comes with navigating something vast and constantly shifting.

Pip charts the non-linear journey many educators take with AI: early excitement, growing concern, and a recognition that the technology can be both a poison and a supplement – a modern pharmakon. This framing underpins NIOT’s stance: teaching is extraordinarily complex, and while AI can support learning, it also risks eroding the “cognitive grapple” trainees need in order to learn.

The session highlights two conditions trainees must have to exercise real agency with AI:
• A firm grasp of cognitive science – understanding how learning moves from working memory into long-term memory
• A practical, ethical understanding of how generative AI works – including its probabilistic nature, biases, and potential unintended consequences

Pip shares concerns that under pressure, trainees may default to cognitive offloading, using AI to think for them rather than with them. Questions of safety, authority, attachment, and the emotional design of chatbots are explored, along with the wider societal risks of misinformation and democratic erosion.

NIOT’s current position includes:
• Encouraging trainees not to use generative AI for formal weekly lesson plans, to protect the developmental function of planning
• Integrating AI education across ITT, ECTP, and NPQs instead of adding extra CPD burden
• Redesigning assessment to avoid offloading (for example, reducing traditional essays)
• Developing a simple five-step checklist for trainees when exploring AI
• Building internal expertise and working collectively on institutional decision-making

Pip closes by emphasising that perfection is not the goal. Transparency about limitations, diverse perspectives, and being problem-led—rather than tool-led—are the principles that must guide AI adoption in teacher education.