Abstract:

The rapid expansion of generative artificial intelligence (GenAI) in higher education has created new opportunities for supporting feedback, dialogue, and student-centered learning. However, many current AI-based tools lack a clear pedagogical foundation and tend to prioritize answer generation over processes that foster self-regulated learning (SRL) and critical reflection.

This paper presents the design of a dialogue-based (Socratic-inspired) generative AI chatbot grounded in empirical evidence from a systematic literature review. The review followed PRISMA guidelines and analysed 110 empirical studies retrieved from Web of Science and Scopus.

The findings show that the use of GenAI in higher education is predominantly concentrated in the self-reflection phase of SRL, particularly in activities related to feedback, evaluation of AI-generated corrections, and reflection on errors. In addition, recurrent patterns of dialogic interaction were identified in tasks involving debate, argumentation, and synthesis during task execution. The results also indicate that the effectiveness of AI-mediated interaction depends on the alignment between task design, metacognitive engagement, and teacher mediation.

Based on these findings, the proposed chatbot design integrates structured questioning, feedback cycles, and SRL-based scaffolding, positioning the system as a cognitive support tool that promotes reflection and active engagement rather than answer provision.

This study contributes an evidence-informed design framework for integrating generative AI chatbots into formative assessment and self-regulated learning processes in higher education.

Keywords:

Generative Artificial Intelligence, Educational Chatbots, Self-Regulated Learning, Formative Assessment, Higher Education.