Abstract
Geography teachers increasingly rely on generative AI for lesson planning, yet outputs frequently exhibit geographical mediocrity: lessons appearing adequate but lacking the spatial thinking and place-based analysis of quality geography education. This risk is amplified for out-of-field teachers lacking disciplinary expertise to recognise AI's shortcomings. Applying the AI-Locus of Control (AI-LOC) framework, we distinguish teachers who surrender pedagogical authority to AI from those retaining agency through disciplinary scaffolding and iterative judgement. A collaborative autoethnographic case study documents one academic's transformation from external to internal AI-LOC. We propose four implementation pathways, contributing an agency-based alternative to competency-based approaches to AI integration.