The Agency Audit: A Practical Framework for Quantifying Human Involvement in AI-Collaborative Tasks

The real risk of AI isn’t a rogue superintelligence deciding humans are obsolete. It’s quieter, more insidious, and already underway—humans voluntarily offloading their thinking, judgment, and communication to AI tools, not through coercion but through convenience, aversion, and the relentless pursuit of efficiency. This session reframes the AI risk conversation for teaching and learning and offers something more urgent and actionable: a practical, research-based framework for assessing human agency in the liminal spaces of Human/AI collaborative work. Drawing on research in AI ethics and learning science, we map the spectrum between AI that augments human thinking and AI that replaces it, and examine why that line is so easy to cross without noticing. The framework is an interactive online tool that assesses three dimensions of agency in Human/AI collaboration. These include Cognitive Agency: are you still doing the thinking, or has the AI done it for you?; Communicative Agency: are your voice and judgment present in the output, or have they been replaced?; and Decisional Agency: are you making real choices, or ratifying what AI has already decided? Participants will leave with tools they can immediately use: a self-assessment instrument for their own AI-assisted workflows; a classroom activity for surfacing student agency in AI use; and design principles for building courses and assignments that protect human cognition without rejecting AI.