| Summary: | This systematic literature review investigates how artificial intelligence enhances
environmental management systems and life cycle assessment practices. By ana
lysing 436 coded concepts from ninety-eight peer-reviewed studies published be
tween 2003 and 2025, the study compares the effectiveness of AI techniques and
identifies best practices for integration into sustainability frameworks. Four key
Application dimensions are revealed: data management and standardization,
explainable AI implementation, multi-objective optimization, and semantic model
ling frameworks.
The findings show that neural networks offer the highest predictive accuracy yet.
face limitations in transparency and interpretability. In contrast, ensemble meth
ods such as Random Forest provide a balanced approach, delivering impressive
performance while supporting explainable, regulator-friendly outcomes. Key success
cess factors include data harmonization strategies that reduce uncertainty by 31%,
the use of explainable AI tools to ensure traceable audit trails for regulatory compliance
pliance, and performance validation techniques achieving over 95% model accuracy
racy with real-time processing capabilities.
Furthermore, strategic optimization methods such as NSGA-II and reinforcement
learning to demonstrate measurable environmental and economic benefits, including
ing up to 16% greenhouse gas reduction and 43% cost savings. This study
provides the first comprehensive, evidence-based synthesis of AI-enhanced EMS and
LCA applications, offering actionable guidance for implementation, particularly
in alignment with Corporate Sustainability Reporting Directive (CSRD), require
ments.
|
|---|