A meta-analysis of computational thinking and artificial intelligence in education: impacts on students' problem-solving skills

Abstrak

Despite the rapid expansion of research on computational thinking (CT) and artificial intelligence (AI) in education, evidence on their comparative effects on students' problem-solving skills remains fragmented and inconsistent, underscoring the need for a systematic quantitative synthesis. This study conducted a systematic meta-analysis to examine and compare the effects of CT-based and AI-based instructional interventions on students' problem-solving performance. Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines, twenty-four empirical studies published between 2017 and 2025 were analyzed using a three-level random-effects meta-analytic model to account for within-study dependencies and heterogeneity. The results showed that CT-based instruction produced a statistically significant and consistent positive effect on problem-solving skills (pooled effect size = 0.30, p < 0.001), indicating high stability across educational contexts. AI-based instructional interventions yielded a larger pooled effect size (0.46, p < 0.001), although greater variability was observed across instructional designs and contexts. These findings suggest that CT strengthens analytical reasoning and systematic problem-solving processes, whereas AI enhances adaptive and reflective thinking through personalized learning support. The study contributes theoretically by clarifying the complementary roles of CT and AI in problem-solving development and practically by providing evidence-based guidance for designing effective technology-enhanced learning environments.