Fifty Years of Topic Development, Thematic Combination, and AI’s Evolving Reach in Computers & Education

Abstract

Across the first fifty years of Computers & Education, successive technologies entered an expanding educational agenda centered on teaching, learning, learner regulation, assessment, participation, access, and institutional conditions. This study examined how that agenda developed from 1976 through partial 2026. We assembled 5,739 records, retained 5,388 eligible articles and reviews, and modeled 5,387 records using titles and available abstracts. An eighteen-topic structural topic model traced topic prevalence across five historical periods. Ordered primary–secondary topic combinations examined changing within-article co-occurrence patterns, while a validated dictionary and continuous piecewise binomial logistic models traced artificial intelligence (AI) prevalence and topical integration. Early volumes concentrated on computer-assisted learning, educational software, and intelligent systems; later research gave greater attention to teacher integration, collaboration, learning analytics, learner regulation, digital literacy, and feedback. Within-article topic combinations diversified substantially across the archive. The share represented by the ten most frequent topic combinations fell from 74.4% in 1976–1989 to 16.5% in 2010–2019, indicating substantial diversification. The human-validated AI corpus contained 390 records (7.2%) and extended across 17 of the 18 topics. Annual modeling identified marked acceleration in AI-related publication prevalence after 2018, as research on generative models expanded alongside established AI traditions. This acceleration reflects growing research attention, not demonstrated educational effectiveness. Across fifty years, the journal broadened how it addressed enduring educational questions. These findings call for evaluating AI within its educational configuration, with attention to human agency, theoretical and disciplinary grounding, and evidence of effectiveness in authentic learning settings.