Rethinking the circular economy in agri-food: human-centred AI for a new circular economy 5.0 paradigm

Purpose – This study offers a systematic literature review on the circular economy (CE), artificial intelligence (AI) and human-in-the-loop (HITL) nexus in the agri-food sector. It specifically investigates how AI-enabled strategies facilitate CE transition within Industry 5.0 contexts, how human roles shape these processes, and how these dynamics vary across for-profit organisations of different sizes. Design/methodology/approach – A systematic review following PRISMA guidelines screened 3,578 records from two databases, selecting 128 articles spanning 2015–2025. Guided by a PSE-informed multi-layer conceptual framework, we analysed variables using a theory-informed protocol guided by three research questions mapping CE transition across AI lifecycle phases, human agency and organisational configurations. Structured content analysis synthesised findings across a series of analytical tables. Findings – The study reveals pronounced imbalances: CE integration tends to remain instrumental and narrow, while systemic uptake is limited across the AI lifecycle, with early phases dominating and HITL configurations are weakly strategical – human roles confined mostly to deployment or monitoring; large enterprises integrate AI within structured ESG strategies, while SMEs face significant barriers linked to cost, skills and infrastructure. Technological efficiency remains the primary driver in research, though motivations associated with systemic CE transition are emerging. Research limitations/implications – The review excludes grey literature and non-English studies. Nonetheless, the developed framework offers a replicable analytical tool for exploring CE 5.0 transitions in agri-food ecosystems.