Artificial Intelligence and the Development of Critical Thinking in Science Education: A Systematic Review

The rapid expansion of artificial intelligence (AI) in education has positioned it as a key driver of digital transformation and changes in learning paradigms. This study systematically examines how AI influences the development of critical thinking in science education, focusing on both underlying mechanisms and instructional practices. Guided by the PRISMA framework, a comprehensive search, screening, and synthesis of literature published between 2021 and 2025 in the Web of Science and Scopus databases were conducted, resulting in a final sample of 21 studies. The findings indicate a sustained growth in research within this field, with mixed-methods approaches predominating. At the technological level, generative AI and adaptive learning systems are the most widely applied, while learning analytics systems, predictive AI, explainable AI, and virtual simulation environments are also gaining increasing attention, forming a complementary and evolving technological ecosystem. At the pedagogical level, AI has become increasingly embedded in instructional design, encompassing teaching models, methods, frameworks, and instructional modules, reflecting a trend toward deeper integration. Despite its potential, the application of AI in educational practice faces several challenges, including issues related to output reliability, risks of student overreliance, and limitations in teachers’ professional competencies. These factors may constrain the effectiveness and sustainability of AI in supporting the development of critical thinking in science education.