Where is the reflexive ‘I’ in the Elements of AI?

AI's opportunities and potential high-risk consequences for individuals and societies render mass AI literacy imperative. MOOCs are one effective conduit for its provision. However, MOOCs remain epistemologically one-sided when lifelong learning steadily shifts towards a reflexive epistemology whereby subjectivities and expert knowledge intersect, problematising the latter's relevance to agents when disregarding the first. Addressing the underexplored epistemologies of AI literacy MOOCs and kindled by transformative learning in late modernity, this paper examines how the design of the MOOC Elements of AI prompts reflexivity over AI. A Bloom's taxonomy-based qualitative content analysis categorised 16 learning objectives and 25 assessments according to cognitive processes and knowledge dimensions they serve. Results showed adequate but delayed instruction for reflexivity and a benign constructive misalignment, with assessment hitting higher and wider processes and dimensions than the learning objectives. Following the fleshing out of results, their discussion leads to EAI-specific and general enhancements for identity-based transformative AI literacy MOOCs catering to scale and individuality.