<![CDATA[The integration of generative AI into academic reading contexts presents both a transformative scaffolding opportunity and a pedagogical risk: without deliberate instructional design, students may default to passive, query-and-paste AI use that reinforces surface-level text processing rather than analytical engagement. This Classroom Action Research (CAR) study investigated the effectiveness of AI-assisted digital annotation as a pedagogical intervention for improving undergraduate EFL students’ critical reading of academic texts. Guided by the Kemmis and McTaggart spiral model, the study was implemented across two iterative cycles with a class of 50 undergraduate students enrolled in Bahasa Inggris course at an Indonesian higher education institution. Data were collected through Critical Reading Performance Tests (pre-test and post-test per cycle) and classroom observation, and were analysed using a mixed-method approach integrating paired-samples t-tests with thematic analysis following Miles, Huberman, and Saldaña’s interactive model. Findings demonstrated statistically significant improvements in critical reading performance across both cycles, with a cumulative mean gain of 28.8 points from pre-test (M = 49.8) to the Cycle II post-test (M = 78.6; p < .001, Cohen’s d = 1.59), and 83.3% of students surpassing the predetermined success threshold. Qualitative findings documented a shift from passive paraphrastic annotation to interrogative critical dialogue with both text and AI tool. The study concluded that AI-assisted digital annotation, when embedded within a structured pedagogical framework and accompanied by explicit instruction in critical AI interrogation, constituted an effective and ecologically valid strategy for developing students’ critical reading competencies in EFL higher education contexts.]]>
Copyrights © 2026
