Artificial Intelligence and Circular Economy: A Systematic Literature Review

Research on Artificial Intelligence (AI) for the circular economy remains fragmented across separate literatures on reuse, recycling, remanufacturing, waste reduction, and resource optimization. Following PRISMA 2020, this review synthesizes 53 studies published between 2019 and 2026, all identified through systematic searches of Scopus and Web of Science; a supplementary Google Scholar search yielded no additional eligible studies. Four AI capability clusters recur across the five strategies: predictive assessment, automated classification, process optimization, and system-level integration. Recycling is the most mature area (16 studies, 30.2%). In this area, computer vision and deep learning sorting systems have been validated in real operating settings, although contamination and inconsistent packaging still reduce accuracy. Waste reduction (11 studies) has a growing base of empirical and pilot evidence, with machine learning forecasting reducing overproduction and food waste, but almost entirely in food and retail supply chains. Resource optimization (7 studies) is concentrated in energy-intensive, data-rich manufacturing, where AI supports energy management and fills data gaps in life cycle assessment. Remanufacturing (6 studies) mainly relies on predictive maintenance and remaining-life estimation, while AI-assisted disassembly remains largely conceptual. Reuse (5 studies) is the least developed strategy: it focuses mainly on condition assessment and reverse logistics and pays little attention to organizational strategy. Across all strategies, adoption depends on data quality, organizational readiness and cost, and regulatory and market pressure. The review proposes a framework that positions AI as a shared, cross-cutting enabler of circular economy transformation. It also sets out a research agenda on field validation, cross-strategy capability transfer, and underexplored sectors and regions.