AI tools like ChatGPT are grabbing headlines, but other AI techniques and tools specifically designed for enterprises are quietly helping companies meet their sustainability goals. Classic AI is already being used widely today in various use cases, and generative AI is evolving rapidly to address new classes of use cases.

I previously led technical teams that helped customers with their AI implementations. When I started a role as a leader for sustainability in Expert Labs, our professional technology services organization, I saw the potential for AI to help with energy efficiency, decarbonization, and waste reduction. Discover the current and emerging use cases for AI in waste management, optimization, energy reduction and ESG reporting.

How AI is helping businesses accelerate their sustainability journey today

• Asset management: Whether it’s for utility infrastructure or factory floor machinery, timely intervention can prolong the life of an asset; reducing the volume of waste sent to landfill and the environmental impact of creating a replacement. AI solutions work by collecting asset performance data and feeding it into machine learning models, which can predict asset health and risk of failure.
• Inventory management: Transportation uses energy;in addition, perishable goods may need to be refrigerated in transit and storage. Inventory optimization is important to ensure you have enough stock while also meeting customer demand. At the same time, you want to reduce the carbon footprint associated with moving and storing stock. AI helps address this problem by combining aspects like demand forecasting, last–mile delivery, and routing optimization.
• Schedule optimization: This use case is like inventory management but addresses the challenge of ensuring that you have the appropriate alignment of talent. If we think about asset maintenance, for example, the questions are which technicians are available, where, and how should their work be prioritized. It’s not about minimizing travel. Instead, it’s better to prioritize a more distant asset for repair because that asset has a higher cost or could fail sooner. AI can tackle problems like asset maintenance efficiently.
• Anomaly detection: Some manufacturers have zero-defect goals. If a part is defective or assembled incorrectly, it might not be possible to salvage or recycle it. Image and video recognition systems can use AI to monitor each stage of manufacture, catching any discrepancies as early as possible. As well as wasted materials, additional energy is consumed when parts have to be reworked or remade. This use case shows how AI can help by processing unstructured image and video data in addition to structured data in the previous examples.
• Compute optimization: Data centers consume a huge amount of electricity.By using AI to understand compute demand over time, it becomes possible to optimize the use of computing and cooling resources.  Matching resources to demand more closely helps to save energy.

Where to next?

Over the next year or so, I expect to see companies deploying generative AI applications that help with a new class of use cases to meet their sustainability goals. Some companies are already working on them.

The first among these is using intelligent document understanding to process sustainability information. Companies use several different frameworks to report their environmental impact in a standardized way. It’s a time-consuming process to collect relevant information and produce ESG reports. Generative AI software retrieves and summarizes text information from various business systems, including supplier systems, and maps it to the reporting frameworks, with the option for human review.

On the other side, AI streamlines processing information already compiled in environmental, social, and corporate governance (ESG) reports. A company could combine purchase order information with a supplier’s ESG report. For example, if you know you’re responsible for half of a supplier’s turnover you can use their ESG reports to estimate your responsibility for scope 3 emissions.

For investors interested in green finance, AI could process ESG reports in bulk to create a recommended shortlist of companies with a stronger environmental posture. In an advanced use case, generative AI models fine-tuned with a company’s sustainability policies could power an advisor application for activities such as supplier selection.

Foundation large language models (LLMs), fine-tuned with domain-specific data, are likely to play an important role in intelligent text processing applications like these.

Foundation models using geospatial data are also likely to make their mark in the coming year or so. These models will be valuable for predicting flood zones, forest fires, and other climate risks. Businesses in sectors including agriculture, retail, utilities, and financial services will be able to use these models for risk assessment and mitigation.

As companies adopt generative AI in these new use cases, they also need to pay attention to a new set of risks that are emerging, ranging from potential privacy concerns to a lack of factuality. A Responsible AI approach and an AI Governance framework are both needed to ensure guardrails are in place for the responsible use of both Classic and generative AI.

Sustainability goals and other business goals go hand in hand. For many of these use cases, there is a close relationship between sustainability and cost. Reducing energy, avoiding waste, and optimizing resources have financial benefits as well as environmental advantages. Using new sustainability applications powered by AI, companies will find it easier to make decisions that are aligned with their sustainability goals.