Redesign assessment reflectively for possible AI use, using supervised tasks for lower-order skills and intentional AI use with reflection for higher-order work¶

Strategy · All strategies
Evidence · no claims cited

Description¶

The chapter recommends adapting or redesigning assessment to account for intentional or incidental AI use, explicitly not relying on GenAI detection software, which has "proven to be inaccurate and unreliable." Student suggestions include in-person essays, oral exams, and presentations with personalized questions. The chapter also describes the AAA (Against, Avoid, Adopt) framework by Lye and Lim (2024): supervise assessment of lower-order skills AI can complete, contextualize higher-order tasks, and design some assessments that use AI intentionally with student reflection.

Design Implications¶

Context¶

Requirements¶

• Assessment methods adapted to account for possible AI (mis)use
• For Adopt-style tasks, students submit their own work plus an evaluation of their AI use

Constraints¶

• Detection software is rejected as inaccurate, unreliable, and risk of instilling a policing dynamic

Target Learners¶

• higher-education students, predominantly engineering

Target Learning Goals¶

• critical thinking and self-reflection about AI use
• authentic demonstration of higher-order skills

Related Strategies¶

• Assess AI competencies continuously with formative quizzes, peer review, reflective journals, real-world summative projects, and portfolios
• Teach subzone classification explicitly as a teachable, assessable cognitive skill in clinical curricula
• Redesign assessment toward supervised, oral, process-based and authentic formats to verify students' own reasoning under GenAI
• Address generative AI in higher education through assessment redesign and proportional reflective practices rather than surveillance or punitive controls
• Graduate AI literacy training should promote purposeful, task-oriented, critically supervised GenAI use rather than targeting overall use levels
• Adapt non-invigilated written assessments and their marking criteria in response to Gen-AI susceptibility
• Use LLM-generated design problems in supervised, time-bound exam-like settings to assess individual higher-order thinking where AI use or teamwork obscures mastery
• Redesign assessments to require critical thinking, originality, interactive activities, and real-time or proctored formats

Examples¶

-

Key Sources¶

• Kudina, O. (2025). Using AI in engineering education: a balancing act, driven by clear purpose. Draft version, to appear in The Routledge Handbook of the Philosophy of Engineering, 2nd ed. https://arxiv.org/abs/2606.16626. doi:10.48550/arxiv.2606.16626