Large Language Models as Socratic Mentors: Transforming Educational Approaches for Epistemic Development
Year: 2025
Type: article
Abstract: This paper examines how Large Language Models (LLMs) can transform educational approaches by functioning as Socratic mentors rather than mere information providers. Drawing on dialogic pedagogy and constructivist learning theories, we propose a framework positioning LLMs as epistemological catalysts that emphasize epistemic agency, perspective multiplicity, and metacognitive development. The implementation of LLMs as Socratic mentors faces challenges including accuracy concerns, bias, transparency issues, and ethical considerations. Comparative analysis with traditional educational methods reveals potential advantages in personalization capabilities, scalability, and engagement potential. Future research should focus on evaluating impacts across diverse learner populations, balancing AI guidance with student autonomy, developing ethical integration frameworks, and investigating long-term epistemic development. This reconceptualization of LLMs as partners in intellectual exploration rather than substitutes for human thinking can enhance cognitive development while preserving uniquely human dimensions of education.
Source: Aracê.
DOI: 10.56238/arev7n5-229
Institutions: Universidade de São Paulo, NACE International
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
SDG: Quality education
Open Access status: bronze
OpenAlex ID: W4410355207