TEACHING READING LITERACY IN THE DIGITAL AGE: FROM TRADITIONAL STRATEGIES TO DIGITAL AND AI-BASED TOOLS

Abstract

Reading literacy in the digital age increasingly requires learners not only to comprehend screen-based and multi-source texts, but also to navigate information environments, evaluate source credibility, and use AI tools critically and responsibly.

The aim of this study is to synthesize research published between 2015 and 2025 on digital reading, online source evaluation, and AI-supported reading instruction, including implications for EFL/EAL contexts. A PRISMA-oriented systematized review was conducted using Scopus, Web of Science, and ERIC, supplemented by a targeted and reproducible Google Scholar search. Eligible sources included peer-reviewed empirical studies, systematic reviews and meta-analyses, and foundational conceptual models addressing reading comprehension in educational settings together with a digital, online, or AI component. The findings were integrated through thematic synthesis. Of 163 identified records, 10 studies met the inclusion criteria. The evidence formed four thematic clusters: strategic digital reading and task-route design; navigation and source evaluation through source prompts, civic online reasoning, and lateral reading; AI-based scaffolding through chatbots and text adaptation; and conditions that shape effectiveness, including digital inequality, ethics, and academic integrity. The analysis indicates that digital and AI tools are pedagogically valuable when they structure learners’ reading strategies, support verification, and preserve independent meaning-making, whereas poorly designed or substitution-oriented use may weaken comprehension and learner autonomy.

The scientific contribution of the study lies in an integrative model linking pre-, during-, and post-reading instruction with bounded digital and AI support.

Practically, the review provides teachers with guidance on designing multi-source tasks, embedding routine source-checking procedures, using safe AI prompts, and reducing risks associated with overreliance on automated assistance. The relatively small and heterogeneous evidence base limits generalization, so the conclusions should be interpreted as converging thematic patterns rather than pooled effect estimates.