<![CDATA[Artificial intelligence (AI) can support academic reading, yet its instructional value depends on whether students use it to interrogate texts or merely obtain ready-made answers. This study developed and preliminarily evaluated an AI-supported Problem-Based Learning (PBL) model designed to promote evidence-based inquiry in EFL academic reading. Following the ADDIE framework, the study produced a six-session instructional module and a structured prompting protocol that guided students to formulate questions, examine textual evidence, evaluate arguments, and reflect on their interpretations. The model was implemented with 22 undergraduate EFL students enrolled in an Academic Reading course at a public university in Indonesia. Data were collected through expert validation, pre- and post-intervention academic reading assessments, a student perception questionnaire, classroom observations, AI-interaction logs, and reflective justification sheets. Quantitative data were analysed using descriptive statistics and a paired-samples t-test, while qualitative data were examined thematically. Students’ mean academic reading score increased from 61.45 to 81.32, with a statistically significant difference between the pretest and posttest scores, t(21) = 9.42, p < .001. The most substantial improvements occurred in coherence evaluation and inferential reasoning. Questionnaire, observational, and interaction-log data also indicated positive student perceptions and progressively greater engagement in inferential questioning, textual justification, argument evaluation, and reflective revision. These findings suggest that structured AI-supported PBL may help shift students’ use of AI from answer seeking toward evidence-based academic inquiry. Nevertheless, the small single-institution sample, absence of a comparison group, and lack of direct cognitive-load measurement require the findings to be interpreted as preliminary.]]>
Copyrights © 2026
