Generative AI in Education: Supporting Learner Autonomy Through Adaptive Interventions

<p dir="ltr">Introduction: Generative AI (Gen AI) has emerged as a transformative force in education, overcoming initial resistance due to concerns about academic integrity. With appropriate safeguards, educators have begun to explore its potential to enhance teaching and learning. This paper investigates the use of a Gen AI chatbot application, AskWhatCher, which integrates the AskCher 3-2-1 exit poll and the AskWhat chatbot to foster self-directed learning (SDL) among students. </p><p dir="ltr">Methods: The study employed a mixed-methods approach, combining pre- and post-usage surveys based on select Motivated Strategies for Learning Questionnaire (MSLQ) scales and qualitative data from focus group discussions (FGDs). A functional chatbot prototype was developed using GPT-3.5-Turbo within a secure, closed ecosystem to mitigate risks such as hallucination and privacy breaches. The chatbot was trained with module-specific content, allowing students to pose their own questions. Exit poll routines scaffolded classroom learning by promoting reflective practices at the end of lessons. The exit poll function allows lecturers to adapt and customize the learning path for individuals with insights generated by AI on the platform. The chatbot provides instant feedback, enabling students to obtain the answers immediately, thus fostering Self-Directed Learning (SDL). Results The integration of quantitative survey results and qualitative FGD insights revealed generally positive trends in key areas of SDL. Students reported increased autonomy and engagement in their learning processes. Quantitative findings demonstrated improvements in SDL dimensions aligned with the Singapore Polytechnic (SP) SDL model. Feedback highlighted the tool's potential to enhance reflection and self-regulation among learners. </p><p dir="ltr">Conclusion: The study's findings suggest that AskWhatCher is a promising tool for fostering SDL. The combination of Gen AI capabilities and structured exit poll routines supports students in developing critical self-directed learning skills. Further iterations and enhancements of the application could lead to greater efficacy in promoting SDL across diverse educational contexts.</p>