Abstract
The advent of Generative Artificial Intelligence (GenAI) transformed technology’s role in the education sector by introducing highly adaptive, interactive tools. Standard implementations are often inadequately designed to effectively aid students. In some cases, these systems hinder genuine learning by inducing cognitive offloading, offering immediate solutions that bypass the cognitive processes necessary for deep conceptual understanding. This concern highlights the need for pedagogically grounded systems that foster genuine understanding and skills development. To address this gap, this paper presents an experimental implementation evaluating the feasibility of embedding formal educational frameworks directly into AI system architecture. We introduce Learning Companion, a prototype engineered to mitigate passive cognitive offloading by translating educational theory into programmatic workflows. Rather than replacing educators or automating learning, the Learning Companion is designed to strengthen meaningful learning interactions by positioning AI as a collaborative partner that supports learner agency through guided dialogue and adaptive instructional scaffolding. Aimed at enhancing critical reasoning and academic writing, the system operationalises Vygotsky’s Zone of Proximal Development (ZPD) through algorithmic Socratic questioning to provide structured instructional scaffolding. This human-centred approach demonstrates how AI can complement teaching and foster deeper learner engagement while preserving meaningful connections between students, educators, and intelligent learning technologies.
Keywords
Human centred learning, socratic learning, Zone of Proximal Development, critical thinking, artificial intelligence.
