Understanding self-directed learning in AI-Assisted writing: A mixed methods study of postsecondary learners

This study investigates how postsecondary learners employ generative AI, specifically ChatGPT, to support their self-directed learning (SDL) for writing purposes. Following a sequential mixed methods design, we analyzed 384 survey responses and 10 semi-structured interviews with postsecondary writers. Findings suggest that the major learning task that the learners used ChatGPT for writing is brainstorming and seeking inspiration for ideas. While the entering motivation for using ChatGPT varies from curiosity about innovative technologies to fulfilling academic requirements, such entering motivation transformed into task motivation when the learners perceived the potential benefits of ChatGPT for assisting their writing. In terms of self-management, participants mostly demonstrated a high responsibility towards their own learning with ChatGPT and employed various strategies for SDL. Although survey respondents demonstrated a comparatively low level of self-monitoring, most interviewees claimed that they critically reflected on their learning process and validated information provided by ChatGPT. There are mixed opinions regarding whether the writing skills have improved as a result of using ChatGPT. Some participants suggested that the benefits brought by ChatGPT, such as alleviating social pressure and receiving instant feedback at any time, encouraged them to spend more time practicing writing and making revisions. However, some argue that assessing their AI-assisted SDL learning progress in the short term is challenging. This study addresses gaps in the existing literature where there is scarce, large-scale empirical research on self-directed AI usage in writing, shedding light on the emerging phenomenon of utilizing generative AI as a means of SDL in writing.