Artificial Intelligence–Driven Carbon Footprint Assessment: A Cross- Sector Review of Methods, Challenges and Future Directions

Abstract

The growing feeling of urgency in regard to the need to reduce climate change has amplified the need to make the calculation of carbon footprint in different industrial sectors more accurate and efficient. The traditional approaches, which include the Life Cycle Assessment (LCA) approach, the Emission Factor Approach, and the Standard Approach, provide a platform upon which the calculation of greenhouse gas emissions can be done. Nevertheless, some recent investigations undertaken on behalf of different industrial sectors, including manufacturing sector, construction sector, logistics sector, transportation sector, and digital infrastructure sector, state that the traditional means of estimating greenhouse gas emissions are limited, including the application of fixed-valued emission factors, rigidity, and high levels of uncertainty in Scope 3 emissions. Meanwhile, Artificial Intelligence (AI) and Machine Learning (ML) techniques are being developed as promising solutions for improving the carbon footprint models. Data science techniques like deep learning networks, ensemble learning, predictive analysis, and IoT monitoring systems can help forecast and monitor the emissions and carbon footprint in real-time. However, the connection between AI-based prediction systems and standardized carbon accounting systems is still fragmented. The research gaps are identified in the areas of inconsistency in boundaries, interoperability, the absence of automation in the calculation of Scope 3, and the need for better modeling of uncertainties. A conceptual framework for the application of AI-based techniques in the calculation of carbon footprint is proposed based on the analysis.India.