Abstract: Currently, the transformative impact of generative artificial intelligence on university students' learning has become a globally recognized focal point. However, the academic community remains divided on the effects of generative artificial intelligence on students' academic performance, with many boundary conditions yet to be explored. A meta-analytic approach is adopted to synthesize 39 experimental studies from internationally renowned English-language journals. Grounded in sociotechnical systems theory, the moderating effects of 15 variables across four key dimensions: sample characteristics, environmental features, material tools, and human-computer interaction, on the relationship between generative artificial intelligence and university students' academic performance. The results indicate that, compared to traditional learning methods, the use of generative artificial intelligence in learning produces a moderate positive effect on university students' academic performance. Moderator analyses reveal that all examined variables exert differential influences on this relationship, with significant moderators primarily concentrated in the dimensions of environmental features and human-computer interaction.Future research and practice should focus on the impact of generative artificial intelligence on students' higher-order cognitive and affective abilities, strengthen the application of technology adaptability in student learning, and develop technology integration models based on distributed cognition principles.