Enhancing Critical Thinking Skills in ChatGPT-Human Interaction: A Scoping Review

Introduction: The rapid integration of generative artificial intelligence (GenAI) technologies, including ChatGPT, into educational environments has introduced both opportunities and challenges for learners and educators. While GenAI can support advanced learning practices, it also raises concerns about critical engagement and the accuracy of generated content. Previous systematic reviews have explored GenAI’s relationship with critical thinking (CT) and self-regulated learning, but a focused synthesis of recent empirical evidence on GenAI’s impact on university students’ CT skills remains lacking. Method: This scoping review followed the PRISMA-ScR guidelines and applied the Arksey and O’Malley framework alongside the Population – Concept – Context (PCC) model. Studies were identified via the Scopus database, using inclusion criteria limited to the years 2024–2025, English language, and the Social Sciences subject area. Thirty eligible empirical studies were analysed and visualised using VOSviewer to identify thematic clusters and categories in the literature. Results: The reviewed studies were grouped into seventeen thematic clusters by the VOSviewer and then manually synthesized into six categories based on semantic interpretation: cognitive and metacognitive development, pedagogical innovation and learning design, academic writing and language learning, AI literacy and learner perception, evaluation and assessment technologies, global and ethical dimensions of GenAI use. The findings were analysed as (1) direct enhancement of CT, (2) metacognitive and reflective gains, (3) contextual factors shaping CT, (4) risks of cognitive offloading, and (5) instructional strategies mediating AI’s effect. 21 publications showed predominantly positive impact of GenAI on CT (idea generation, conceptual understanding, construction of arguments, literature review, academic writing, etc.) whereas reported found mixed impact. Conclusion: The review concludes that GenAI holds substantial potential to support CT development, particularly when pedagogically integrated to promote active reasoning, metacognitive monitoring, and critical autonomy. However, the evidence base is still emerging and is limited by its short temporal scope, narrow database coverage, and reliance on self-reported data. Future research should focus on long-term effects, discipline-specific instructional models, and robust theoretical frameworks linking AI use to cognitive outcomes.