Abstract:

This study introduces Socratic GenAI, a reasoning-oriented generative AI framework designed to enhance critical and computational thinking, self-efficacy, and programming performance. Grounded in constructivist, Socratic, and social cognitive theories, the framework redefines AI – learner interaction from answer delivery to guided questioning, encouraging reflection, reasoning, and learner agency. A randomized controlled field experiment will be conducted among programming students at the University of Kinshasa to compare Socratic GenAI with conventional generative AI systems such as ChatGPT, Copilot, and Gemini. Quantitative analysis using structural equation modeling (SEM) will test the mediating effects of reasoning and self-efficacy on learning outcomes.

Expected results suggest that Socratic GenAI substantially strengthens conceptual reasoning and motivation, demonstrating that pedagogical alignment – rather than technological sophistication – determines AI’s educational value. The study will provide empirical grounding for integrating question-centric AI design into computer science education, especially in resource-limited contexts. Beyond theory, it highlights a scalable and culturally adaptive approach to digital tutoring that bridges both cognitive and infrastructural divides. The findings are expected to inform AI pedagogy, education policy, and EdTech entrepreneurship across Sub-Saharan Africa, positioning AI not merely as a productivity tool but as a partner in cultivating human reasoning.

Keywords:

Socratic GenAI, Computational Thinking, Self-Efficacy, Constructivist Pedagogy, Programming Education, Sub-Saharan Africa.