AI-powered sustainability: A new era for Scope 3 emissions management

Introduction

Scope 3 emissions are becoming a critical focus for businesses to reduce their environmental impact. Scope 3 emissions cover a business's indirect emissions from non-owned or controlled assets and are often the scope with the highest environmental impact. These emissions include the entire supply chain and every stage of a product's lifecycle. A company must include the stages of a Life Cycle Assessment (LCA) to accurately understand their product’s emissions which includes: raw material extraction, manufacturing and processing, transportation, consumer usage, and end of life disposal.  

LCAs can enhance Scope 3 insights

An LCA is a tool that provides a comprehensive approach to understanding a product’s environmental impact. LCAs help companies identify key areas of action to reduce their environmental impact and meet sustainability goals. The GHG Protocol and ISO standards dictate that companies should use LCAs as the standard approach for measuring the environmental impact of consumer packaged goods. Traditionally, companies find LCAs to be time-consuming and expensive, but AI now makes Scope 3 reporting more efficient and effective, allowing companies to focus on actions to reduce their impact rather than the calculation itself. 

Challenges to performing an LCA

Companies traditionally perform LCAs manually, requiring significant time and resources from employees. Companies spend a lot of that time finding data and performing research and analysis, often making this step the most time-consuming and difficult part of performing an LCA. While companies collect detailed upfront data, many gaps remain because they need to match the activities in the product lifecycle to specific emissions factors that are not always available or intuitive. This typically requires a skilled understanding of the LCA process and regional impacts for various emissions factors. 

This exercise often involves researching a variety of activities involved in the manufacturing process of the product and that of inputs in the supply chain to be able to compute their environmental impacts. For example, if there are no matches for emission factors for a specific input chemical, it would involve researching the steps involved in the manufacturing of the chemical to find a close enough proxy for it or breaking down the steps of manufacturing and computing the emissions involved for each step. 

A comprehensive dataset allows LCAs to be most effective, ensuring the most accurate results for a product. 

The role of AI in revolutionizing LCAs

Recent advances in AI, particularly Generative AI, have revolutionized various aspects of our lives, from creative writing to complex task automation across fields like healthcare, education, IT, and entertainment.

This has been made possible by the development of large language models by organizations such as OpenAI (ChatGPT) and others. On top of this, a lot of new methods and technologies are being developed to make effective use of these models across a variety of new use cases. One example of a new generation of applications includes perplexity.ai, that helps people find answers to somewhat complex questions. It does so by analyzing the different "web search" queries that are needed to answer the question, gathering the web pages corresponding to those queries, and then performing an analysis grounded on that data to give the final answer. 

One use case of AI in LCAs is data prediction. It fills missing information during the calculation phase, leveraging vast existing datasets, ensuring accurate and reliable assessments.

Impact Factors: Companies use impact factors to quantify the different types of environmental impacts a product or activity has. Examples include, but are not limited to, global warming, ozone depletion, land use, and acidification. Activities throughout a product’s life cycle have different effects on each of the impact factors which requires data for each stage of a life cycle to accurately assess the impact of a product.

Obtaining primary data for upstream activities, like third-party transportation emissions or specific ingredients, is challenging. Impact factor databases help fill these gaps but are sometimes incomplete. AI can automate and refine the process of finding proxies or breaking down activities into smaller steps, providing the necessary data more efficiently.

Leveraging AI to model complex subcomponent data can enhance the quality and completeness of the dataset, leading to more accurate assessments. 

Supply Chain Data: Supply chain data, like production emissions, is essential for an accurate LCA. However, suppliers may lack the necessary data or be unable to provide it which is often a barrier in data collection. This data varies by supplier based on the products and services they offer, underscoring the importance of using data that accurately reflects each supplier’s activities. 

For example, supplier location plays a role in emissions production due to differences in infrastructure that can affect environmental impacts, such as emissions from energy consumption. Additionally, companies must consider transportation routes, as predicting these accurately influences data reliability. Addressing these gaps is crucial for enhancing LCA accuracy and reliability.

Product Data: Product-level data is critical for conducting an LCA which enables more effective strategies for environmental reduction efforts. For example, a product may contain a complex subcomponent which requires data for its constituent parts. This data may not be easily available, but the use of AI can help model what the subcomponent is made of and provide this necessary information. Leveraging AI to fill data gaps allows companies to focus their efforts on specific and impactful actions to reduce their environmental impact. 

Regional and Geographic Factors: Geography can drastically affect the environmental impact of a product. Standards for electricity and water, manufacturing methods, as well as socioeconomic factors, can cause one country’s emissions factor for an activity to be significantly different than for another country. Companies must ensure that emissions factors and data accurately reflect the geographic region of activity to achieve accurate results and identify impactful actions. 

Other Benefits:

Focused Data Collection: Filling in data that can be identified by AI allows companies to focus primary data collection efforts on areas that matter most.

Quicker Results: AI streamlines LCA results, reducing the time required to receive results across many products.

Great Portfolio Coverage: AI increases the coverage of product portfolios, which enhances accuracy and provides a holistic picture of environmental impact.

Quality Control: AI predictive models can help ensure that primary data is accurate and reliable by identifying outliers or other data that needs more scrutiny. 

AI-driven recommendations for reducing environmental impact

Once the LCA results are available, AI provides actionable recommendations to reduce environmental impact. These AI-driven insights help businesses identify and implement effective strategies to lower their Scope 3 emissions, enhancing overall sustainability efforts. Additionally, AI considers practical use cases alongside emissions reduction.

For example, while a material such as glass might be more sustainable, it’s not ideal for a shampoo bottle due to the risk of breaking. AI can research existing publications to find suitable alternatives and evaluate their environmental footprints, shortlisting the best options. This process ensures that the materials chosen not only reduce emissions but also meet the practical requirements of the product's use case.

By leveraging AI in this way, businesses can make informed decisions that balance sustainability with functionality, leading to more effective and practical environmental strategies.

Looking ahead

Recent advances in AI like Generative AI and Large Language Models have made the integration of AI into Life Cycle Assessments (LCAs) possible, marking a transformative advancement in managing Scope 3 emissions. By addressing traditional challenges like time-consuming data collection, high costs, and data accuracy, AI-powered LCAs provide a more efficient, precise, and cost-effective solution. AI enhances data prediction, fills critical gaps, customizes assessments to regional and supplier-specific contexts, and ensures higher quality control. These improvements allow businesses to gain deeper insights into their supply chains and implement targeted strategies to reduce their environmental footprint.

AI-powered LCAs are not just a tool for compliance but a strategic asset for sustainability. Companies can achieve the benefits of LCAs through traditional approaches or AI; however, AI allows them to perform LCAs across their supply chains in a cost effective manner for the first time ever, while maintaining consistency and accuracy.