Conditional Effects of AI Homework Tools on Students’ Academic Performance: A Systematic Synthesis of Empirical Evidence

Abstract

The rapid diffusion of generative artificial intelligence (AI) tools into educational contexts has fundamentally transformed how students approach homework, academic writing, and independent learning tasks. Whilst AI-assisted homework tools promise efficiency, personalization, and immediate feedback, there remains some debate over their implications for academic performance and learning quality. The present study proffers a thorough synthesis of empirical evidence, examining how students' academic performance differs when using AI homework tools compared to traditional homework methods. The review draws on experimental, quasi-experimental, and observational research conducted across secondary and higher education contexts. The findings of the study indicate that AI homework tools are associated with significantly higher grades and writing scores in most controlled comparisons, particularly in language learning contexts, with effect sizes ranging from medium to large. However, the evidence also reveals important trade-offs, including reduced knowledge retention, lower originality, and diminished critical thinking in some settings. The synthesis demonstrates that AI tools primarily optimize output quality rather than learning processes, and that their effectiveness is highly conditional on task characteristics, assessment timing, implementation fidelity, and learner characteristics.