Metacognitive regulation and ChatGPT-assisted writing: introducing an interactive model for cultivating strategic engagement

Abstract

Given the contemporary global issues and growth of generative artificial intelligence (GenAI) technology, students are required to effectively activate their metacognitive strategies when using AI not only to develop essential 21st-century skills but also to leverage the benefits of GenAI and avoid its potential negative effect. This paper explores the dynamic realm of AI-assisted writing within the framework of metacognitive regulation (MR) and self-directed learning. The study introduces a data-driven model of MR derived from examining participants’ use of MR strategies when employing ChatGPT for academic writing. Adopting an in-depth qualitative approach, semi-structured interviews were conducted with 25 postgraduate students who were purposively selected from three distinct Moroccan universities. Based on reflexive thematic analysis, findings revealed that students with higher MR were found to regulate the use of ChatGPT by aiding them in various aspects, including planning, monitoring, revising, information management, and debugging during their academic writing. Based on the findings, a contextualised metacognitive regulation framework for AI-assisted L3 writing was proposed to explain the interaction between MR and ChatGPT within AI-assisted writing. At the core of this framework is the recognition of MR and self-regulated learning. It posits that, for learners to achieve a higher level of MR, a collaborative effort is required to foster agency and autonomy by effectively leveraging the features and affordances of ChatGPT. The study not only sheds light on the theoretical foundation of MR but also provides empirical findings, offering practical implications for cultivating MR in AI-assisted writing contexts.