AI-interaction literacy: reflections on how generative AI might be used to support self-regulated learning in higher education

This reflective paper explores how Generative Artificial Intelligence (GenAI) can function as a conversational partner for self-regulated learning in higher education, and how such use complicates assessment. Using an illustrative interaction with a GenAI chatbot based on a take-home examination question from cognitive science, ‘Do machines think?’, we apply the Structure of observed learning outcome (SOLO) taxonomy, to analyse qualitative differences in the structural complexity of the interaction itself. Our demonstration reveals that GenAI’s default output consistently remained at the quantitative, multistructural phase, producing polished but pedagogically thin responses, and that reaching the qualitative phase required repeated meta-level interventions from the learner, including requests for simplification and structured learning support. Critically, this process demanded the very self-regulatory skills the tool was expected to support. We introduce the concept of AI-interaction literacy to describe the capacity to steer, evaluate, and learn from iterative GenAI interaction, and argue that unguided use may disproportionately benefit already advantaged students. We conclude with three propositions: students should treat GenAI as a critical dialogue partner; teachers should explicitly teach AI-interaction literacy with attention to equity; and take-home assessments should emphasise evaluating knowledge at the qualitative phase of the SOLO taxonomy.