Rewrite The effect of ChatGPT on students’ learning performance, learning perception, and higher-order thinking: insights from a meta-analysis as a headline for a science magazine post, using no more than 8 words

Conclusion

This study used meta-analysis to analyze the impact of ChatGPT on student learning performance, learning perception, and higher-order thinking. With regard to learning performance, an analysis of data from 44 experimental and quasi-experimental studies showed that the calculated effect size indicates a large positive impact (g = 0.867) of ChatGPT on student learning performance. The analysis of moderating variables revealed no significant differences in the effect of ChatGPT across grade levels, role of ChatGPT, or area of ChatGPT application. However, significant differences were found in the type of course, learning model, and duration.

For learning perception, an analysis of data from 19 experimental and quasi-experimental studies revealed a medium positive effect (g = 0.456) of ChatGPT on student learning perception. The moderator analysis indicated that there were no significant differences in its impact across grade level, type of course, learning model, role of ChatGPT, or area of ChatGPT application. However, duration showed a significant effect.

In terms of higher-order thinking, data from nine experimental and quasi-experimental studies indicated a medium positive effect (g = 0.457) of ChatGPT on students’ higher-order thinking. The analysis of moderating variables revealed no significant differences in the effect of ChatGPT across the learning model, duration, or area of the ChatGPT application. However, significant differences were observed in the type of course and role of ChatGPT.

In summary, the meta-analysis results of this study confirm the positive impacts of ChatGPT on learning performance, learning perception, and higher-order thinking, while highlighting some variations in effectiveness across different usage conditions and learning contexts. Based on these results, governments could introduce policies to promote the integration of ChatGPT into education, encouraging universities and secondary schools to apply ChatGPT appropriately. Additionally, ChatGPT could be actively incorporated into STEM and related courses, skills and competencies development training programs, and language learning and academic writing courses to enhance student learning.

Implications for theory and practice

This meta-analysis makes significant contributions to our understanding of the theoretical and practical aspects of the use of ChatGPT in education. From a theoretical perspective, it extends the existing literature on ChatGPT and student learning by providing quantitative evidence that ChatGPT can indeed effectively improve learning performance, enhance learning perception, and foster higher-order thinking. Second, this study provides a more thorough and integrated investigation of potential moderating variables. It explores the effects of ChatGPT on student learning across different types of courses, grade levels, learning models, intervention durations, instructional roles, and application types. It has a broader scope than the meta-analyses by Heung and Chiu (2025) and Guan et al. (2024). Compared to Deng et al. (2025), this study further explores learning methods, the role of ChatGPT in learning, and its application across various domains. Its findings reveal that different learning methods and the roles played by ChatGPT significantly affect students’ learning performance and higher-order thinking, resulting in notable differences in outcomes. Broadly speaking, this study provides a theoretical foundation for the future development of national AI education policies and the application of ChatGPT by educators, while further enriching the research on the use of ChatGPT in the field of education.

From a practical perspective, the results of the analysis of potential moderating variables in this study show that ChatGPT should not be implemented arbitrarily. Instead, it should be used scientifically and reasonably according to the type of course, learning model, duration, etc. Future applications of ChatGPT should pay attention to the following points: (1) When applying ChatGPT to developing students’ higher-order thinking, it is crucial to provide corresponding learning scaffolds or educational frameworks (e.g., Bloom’s taxonomy) because ChatGPT lacks creativity and critical thinking. Doing so can improve its effectiveness in cultivating higher-order thinking. (2) Broad cross-grade use should be encouraged to support student learning. ChatGPT can serve as an auxiliary learning tool for both middle school and university students, enhancing both their academic performance and their learning perception. In middle schools, teachers can use ChatGPT to maintain students’ interest while improving their understanding and memory of complex concepts. In universities, teachers can guide students to actively use ChatGPT in relevant courses to provide personalized learning suggestions to improve their learning. (3) ChatGPT can be actively applied in different types of courses to enhance student learning. For example, in STEM and related courses, ChatGPT can be used to present diverse information to help students think divergently and complete projects. Alternatively, ChatGPT can be used to simulate problem-based learning environments by generating scenario-based questions. In skills and competencies development training, ChatGPT can be used to provide more targeted guidance. In language learning and academic writing courses, ChatGPT can help students enhance their comprehension of texts and improve the accuracy and grammatical correctness of their writing. (4) ChatGPT can be applied in different learning models, particularly in problem-based learning, as its use in this mode best supports students’ learning performance. (5) Continuous use of ChatGPT can be ensured to support student learning, with a recommended duration of four to eight weeks. For shorter durations, teachers should provide relevant scaffolding, such as guidance on how to write high-quality ChatGPT prompts. If a longer duration is necessary, teachers can consider introducing new instructional models, such as blended learning or flipped classrooms, and apply ChatGPT more strategically in different teaching stages such as pre-class preparation, in-class interaction, or post-class review to enhance learning outcomes. Moreover, teachers can conduct regular formative assessments to monitor students’ progress and identify gaps in their understanding. This would help prevent the students from developing an over-reliance on technology that could hinder their knowledge retention, while also promoting deeper learning and long-term mastery. (6) ChatGPT should be used flexibly in teaching as an intelligent tutor, partner, learning tool, and more. Its role(s) can be adjusted according to the requirements of the teaching and learning activities to which it is applied.