Aligning AI affordances with critical thinking skills in Chinese EFL: a discrepancy-based needs analysis for an AI-integrated blended reading module

Introduction: Artificial intelligence (AI) is increasingly embedded in blended learning (BL) environments, yet its integration frequently emphasizes efficiency and performance outcomes rather than the systematic cultivation of higher-order cognition. Addressing this gap, the present study employs a discrepancy-based needs analysis to investigate how critical thinking skills (CTS) can be deliberately aligned with AI affordances in blended reading among Chinese undergraduates. Methods: Drawing on qualitative evidence from 20 Chinese EFL undergraduates, this study collected data through AI-mediated reading interactions, reflective learning logs, semi-structured interviews, and a focused literature-technology scan. A discrepancy-based needs analysis was conducted to examine the alignment between AI affordances, instructional design, and critical thinking development. Results: The findings suggest three interrelated structural misalignments between AI affordances, instructional design, and critical thinking development: (1) predominance of literal comprehension over analytical and evaluative processing, (2) limited metacognitive self-regulation, and (3) instrumental use of AI for surface-level linguistic assistance rather than structured reasoning support. Despite these patterns, learners appear ready for CTS-oriented blended learning and articulate clear expectations for explicit cognitive modelling, dialogic engagement, and adaptive AI feedback that evaluates reasoning quality rather than correctness alone. Discussion: In response to these findings, the study proposes a design-oriented Technology-CTS Needs Matrix that integrates empirical findings from participant data with literature- and theory-informed design extrapolations, systematically mapping Facione's six CTS dimensions onto targeted AI and pedagogical affordances. By reconceptualising needs analysis as a mechanism for instructional alignment rather than deficiency identification, the study provides a theoretically integrated and empirically grounded foundation for structuring AI-mediated blended reading environments that foster critical thinking skills in EFL contexts.