This Action Lab invites participants to examine cognitive offloading as a critical capability in human-AI collaboration for learning and teaching. In everyday academic life, LLMs can reduce overload, support drafting, structure ideas, and make complex tasks feel manageable. Yet many educators also sense a more uncomfortable pattern: we may be over-consuming AI support because it offers quick comfort, speed, and relief, even when the task still belongs to our own professional judgement. The lab therefore asks a high-stakes question: when does cognitive offloading strengthen educators, and when does it quietly erode confidence, responsibility, and agency?

The session links directly to the EDEN 2026 theme by focusing on the human, relational, and ethical dimensions of AI use. It also connects with an EdTech Talents knowledge service on higher education faculty upskilling for meaningful LLM use in teaching and learning. A recent class experiment provides one starting point: students were asked to use GPT to understand what was required to pass a course, and their responses revealed that AI can help surface ambiguity in course design, attendance rules, and assessment criteria, but it cannot replace clear pedagogical responsibility or human interpretation. AI becomes not only an assistant, but also an audit lens.

At the same time, a revealing trend is emerging. Students may use AI extensively themselves, yet they still tell teachers, very explicitly, that in the classroom they want to see what the teacher knows without AI. They have a right to expect this. The question is not whether teachers use AI at all, but what unique human value remains visible, trusted, and worth not replacing. The lab will therefore ask what teachers should deliberately keep as human cognitive work, which tasks can be responsibly offloaded, and which should remain under clear human control when student data, care, or ethical judgement are involved. This includes limits around student-facing communication, where even routine emails may contain personal data.

The moderator will also bring a current practice example: because written feedback can now be questioned as authentically human, she has shifted toward oral face-to-face feedback, while still using short, clearly marked avatar videos as supplemental teaching material. Real is real, artificial is artificial, and both can have a place when their role is transparent.

The purpose of the session is not to shame AI use, but to restore intentionality, dignity, and self-confidence. By the end of the lab, participants will have mapped their own patterns of helpful and harmful offloading, reflected on the possibility of AI detox, and identified practical ways to reclaim 4 to 5 hours per week for rest, reflection, creativity, or richer human interaction.