Students Know AI Should Not Replace Thinking, but How Do They Regulate It? The TACO Framework for Human-AI Cognitive Partnership

A new study from Cecilia Ka Yuk Chan, published on arXiv, tackles a critical issue in AI education: students know they shouldn't let AI replace their thinking, but they struggle to regulate that boundary in practice. Based on data from Hong Kong secondary students, the research found that awareness of ethical AI use did not consistently translate into structured regulation. Students conceptually endorsed the idea that AI should not substitute for their own cognition, yet this belief failed to guide their actual behavior when using generative AI tools. The study highlights a disconnect between ethical belief and strategic execution, and between conceptual endorsement and operational behavior.

To address this gap, Chan introduces the TACO framework (Think-Ask-Check-Own), a process-oriented model designed to help students operationalize the boundary between cognitive support and cognitive substitution. The framework guides learners through four stages: Think (engage with the problem independently), Ask (use AI for assistance), Check (verify AI outputs), and Own (integrate insights into personal understanding). By shifting the focus from teaching students that AI shouldn't replace thinking—which they already know—to providing them with concrete mechanisms for cognitive regulation, the study offers a practical tool for educators. The TACO framework aims to sustain AI as a dynamic cognitive partner in education rather than a crutch that undermines learning.