From AI Capability to Sustainable Supply Chain Performance: A Theory–Method–Context Review

Generative artificial intelligence (GenAI) has prompted claims of improved supply chain performance. This structured critical Theory–Method–Context review examines 24 unique studies published between January 2024 and 28 August 2026 across 15 outlets, distinguishing organisational practices from economic, environmental, and social outcomes. The bounded compilation contains 19 questionnaire studies, two conceptual contributions, one expert-priority study, one case/simulation study, and one archival panel study. Source verification identifies one survey with explicit predictor–outcome temporal separation and another reporting two waves without sufficient detail to establish that separation. The evidence is geographically concentrated and uses heterogeneous AI constructs. A corrected catalogue of 20 selected coefficients and four within-study comparisons illustrates, but does not establish, stronger associations with intermediate mechanisms. One matched comparison is nearly equal, and sensitivity scenarios show that interpretation depends on differential method inflation. The review develops a conditional explanation linking resources, practice change, verification, and outcomes, with an evidence map, six gaps, and eight testable propositions. Net sustainability benefits remain unverified without explicit environmental and social measures, supply chain boundaries, and AI resource accounting. Selection and independent-coding records are incomplete, limiting reproducibility and generalisation.