Scaffolding classical literacy: Reconciling artificial intelligence with conventional reading pedagogy in digital secondary and higher education
Year: 2026
Type: article
Abstract: Background: The rapid integration of artificial intelligence (AI) in secondary and tertiary education has generated concerns regarding cognitive atrophy, particularly the erosion of deep reading and canonical literacy. Aim: This paper investigates how artificial intelligence can be strategically leveraged within digital institutional frameworks not to replace, but to rehabilitate and augment conventional pedagogies, specifically silent text-based comprehension, sustained reading, and classical humanistic scholarship. Approach: Drawing on cognitive load theory, the deep reading architecture, and neo-Vygotskian scaffolding paradigms, this narrative synthesis critically examines empirical and theoretical literature surrounding generative natural language interfaces and adaptive reading environments. Findings: Current evidence reveals that when AI serves as an interactive Socratic interlocutor, lexical glossator, or contextual dialectic partner, it counteracts the superficiality of hyper-reading. Rather than bypassing textual engagement through extractive summarisation, pedagogically constrained AI systems foster structural text comprehension, inferential reasoning, and classical hermeneutic analysis. However, unconstrained automated summarisation poses significant risks of cognitive offloading and superficial text processing. Implications: Artificial intelligence must be intentionally designed with pedagogical friction to reinforce conventional textual scholarship. Higher education and secondary curricula must treat algorithmic tools as cognitive scaffolds that preserve the neurocognitive benefits of deep reading rather than transactional transaction engines.
Author: LCAMH Faculty
Language: English
Subfield: Education
Field: Social Sciences
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7233567006
FWCI:
Cites:
Cited by:
Related to:
