Abstract
The rapid integration of generative artificial intelligence (GenAI) into education has generated both optimism regarding its pedagogical potential and concern about students’ increasing reliance on AI-generated content. This paper critically examines the conditions under which ChatGPT may contribute to the development of higher-order thinking skills (HOTS) in English as a Foreign Language (EFL) education. The study employs a qualitative critical literature review to synthesise empirical studies, systematic reviews, meta-analyses, and relevant theoretical literature concerning ChatGPT, higher-order thinking, critical thinking, cognitive engagement, and cognitive offloading. The reviewed evidence is interpreted through the analysing, evaluating, and creating dimensions of the revised Bloom’s taxonomy. The synthesis indicates that ChatGPT can provide meaningful cognitive and linguistic support when learners are required to question, compare, verify, evaluate, and transform AI-generated information. However, such benefits are not inherent in the technology itself. Rather, they depend substantially on instructional design, teacher guidance, and the extent to which learners retain responsibility for essential cognitive processes. The review also identifies cognitive offloading and uncritical acceptance of AI-generated content as important risks when ChatGPT is used primarily as an answer-generating tool. Overall, the evidence suggests that ChatGPT is most pedagogically valuable when positioned as a dialogic and cognitive scaffold rather than as a substitute for learner thinking. The paper therefore proposes a conditional perspective in which the relationship between ChatGPT and HOTS is mediated by instructional design and learner cognitive engagement. This perspective has particular relevance for EFL pedagogy because AI-supported language learning can be designed to develop linguistic competence while simultaneously strengthening analysis, evaluation, creation, and reflective inquiry.