Large language models like ChatGPT are beginning to reshape classroom dynamics. This study, led by Jin Wang and Wenxiang Fan at Hangzhou Normal University and published in Humanities and Social Sciences Communications, cuts through the noise by offering a balanced, data‑driven view of AI’s impact on student learning.
Drawing on 51 research papers from November 2022 to February 2025, the analysis reveals that students using ChatGPT tend to achieve better academic results. The tool also gives a modest boost to learning perception, suggesting that—even amid concerns over misuse—it enriches the overall educational experience. The meta‑analysis further indicates positive, albeit subtle, gains in higher‑order thinking skills like problem‑solving and creativity, with courses in STEM and practical skill areas reaping the most benefit.
The study also highlights the importance of the instructional model. For example, problem‑based learning environments seem to enhance the benefits associated with ChatGPT, and longer engagement with the tool is linked to stronger outcomes. For anyone weighing the benefits of AI integration, such insights can help in designing effective, balanced learning experiences.
While the paper focuses primarily on college settings and acknowledges the rapid evolution of LLMs, it offers valuable guidance for educators and developers alike. The call is clear: blend the convenience of direct information delivery with strategies that encourage deeper, critical thinking.
