The ease of access to Generative AI in education presents a paradox: it accelerates information retrieval but risks triggering cognitive offloading, which weakens Self-Regulated Learning (SRL). In order to restore student learning autonomy, this study will examine recent research on the possible incorporation of the Socratic Questioning approach into AI systems. 23 chosen papers (2020–2025) from the Scopus and SINTA databases are subjected to a Systematic Literature Review (SLR) procedure utilizing the PRISMA protocol. Thematic analysis results indicate that standard AI configurations tend to reduce metacognitive abilities. Conversely, intelligent agents based on Socratic logic are proven effective in stimulating critical reasoning, yet their implementation is dominated by complex web platforms. A significant gap exists regarding the adaptability of this method on low-resource platforms like WhatsApp, which are more inclusive. In order to democratize access to adaptive virtual tutors, the study's conclusion suggests creating a hybrid architecture that connects Socratic pedagogical intelligence with instant messaging accessibility.
Copyrights © 2025