FROM PREDICTION TO PERSONALIZATION: MAPPING THE EVOLUTION OF AI AND MACHINE LEARNING IN E-COMMERCE CONSUMER BEHAVIOR RESEARCH

Digital transformation has expanded the use of artificial intelligence (AI), machine learning (ML), data analytics, recommendation systems, and personalization in e-commerce. These developments create a need to understand not only the technical capabilities of AI but also how the technology influences consumer behavior, purchasing decisions, satisfaction, engagement, and loyalty, as well as issues regarding trust and privacy. This review paper aims to map the evolution of research on AI and consumer behavior in e-commerce by conducting a content analysis of 316 documents indexed in the Scopus database. The analysis examined publication years, document types, keywords, themes, methodological approaches, citations, and findings derived from titles and abstracts. The results indicate that 274 documents (86.71%) were categorized under the theme of consumer behavior/purchasing, while 256 documents (81.01%) fell under the theme of AI/machine learning. Regarding methodology, 247 documents (78.16%) utilized machine learning or AI models, 132 (41.77%) employed or developed frameworks or models, 87 (27.53%) used data mining or analytics, and 67 (21.20%) applied regression or econometric analysis. Publication growth was particularly strong during the 2024–2026 period, accounting for 276 documents (87.34% of the corpus). The synthesis reveals four main research streams: (1) consumer behavior prediction and modeling, (2) personalization and recommendation systems, (3) marketing analytics and purchasing decisions, and (4) trust, privacy, transparency, and ethical AI. Recent literature indicates a shift from a primary focus on improving prediction accuracy toward issues such as human-centered AI, explainability, user control, governance, and the gap between experimental model performance and real-world marketplace implementation. This paper concludes that research on AI and consumer behavior in e-commerce is evolving rapidly but remains conceptually and methodologically fragmented. Future research agendas need to strengthen longitudinal studies, field experiments, mixed-methods approaches, and cross-cultural comparisons, as well as the evaluation of ethical and governance impacts.