Abstract:

Significance In the context of China's 'dual-carbon' targets and the rapid restructuring of global green trade rules, the textile industry is under increasing pressure to advance green transformation. Artificial intelligence (AI) offers potential for pollution and carbon reduction across different stages and scenarios. However, existing studies remain fragmented, mainly focusing on single stages or individual technologies, and lack a full life-cycle analytical framework. This study reviews AI-enabled decarbonization technologies, identifies innovation pathways, and compares applications across design, production, distribution, consumption, and recycling from a textile life-cycle perspective. By linking AI applications with digital product passport (DPP) data, this study clarifies pathways for coordinated value-chain decarbonization and provides theoretical and practical references for the low-carbon transformation of the textile industry.

Progress AI applications for pollution and carbon reduction now extend across the textile life cycle, but their maturity, depth of implementation, and degree of cross-stage integration vary substantially. At the design stage, generative AI and intelligent decision-making systems support a shift from mass production and excess inventory toward demand-driven design and more precise product matching, thereby reducing material waste and overproduction at the source. However, limited technical standardization and restricted data access still constrain wider applications. At the production stage, deep-learning-based visual inspection and intelligent process control have been relatively well developed in quality control, energy-efficiency improvement, and cleaner production, making production the most mature field of AI-enabled decarbonization in textiles. The main barriers are standardized industrial data collection and the integration of heterogeneous data from different production systems. At the distribution stage, AI-based demand forecasting and supply-chain coordination can reduce inventory accumulation and unnecessary production, but full-chain optimization is limited by insufficient data sharing among firms and by commercial barriers. At the consumption stage, AI recommendation systems and virtual try-on technologies can indirectly reduce emissions by improving product-consumer matching and extending garment use, although algorithmic ethics and user privacy require clearer governance. At the recycling stage, intelligent sorting and digital traceability improve fiber identification and quality control, supporting the transition from end-of-life disposal to higher-value recycling. Yet high equipment costs and unstable markets for recycled fibers continue to restrict large-scale deployment. PP provides a data infrastructure for connecting life-cycle information across stages and actors, offering a basis for coordinated decarbonization beyond isolated technological improvements.

Conclusion and Prospect AI has established an initial full life-cycle enabling framework for the textile sector, providing a practical basis for moving from isolated energy-saving measures toward systematic pollution reduction and decarbonization. However, the full mitigation potential of AI remains constrained by algorithmic opacity, limited model generalizability, unaccounted energy consumption from AI computation, and insufficient cross-enterprise data coordination. Future research should focus on four key directions. Firstly, explainable AI should be further applied to textile process optimization to improve engineer intervention while maintaining model accuracy. Secondly, federated learning-based data collaboration mechanisms should be developed to address industry data silos while protecting data ownership and commercial privacy. Thirdly, DPP data standards and pathways for international alignment should be strengthened to support trusted carbon footprint transmission and dynamic accounting across stakeholders. Fourthly, AI-based multi-objective decision-making systems should be developed to jointly optimize carbon emissions, chemical oxygen demand, energy efficiency, and product quality.

Key words: artificial intelligence, textile industry, full life cycle, pollution and carbon reduction, digital product passport