How generative AI can help the energy industry improve efficiency

by Casey Werth , GM for IBM 's Global Energy Industry and Bryan Sacks , CTO for IBM's Global Energy Industry.

As organizations of every kind look to meet their sustainability goals, the implications for energy, environment, and utilities companies are particularly acute: according to some estimates, the shift to clean electrification could potentially eliminate 70% of the world’s greenhouse gas emissions.[1] AI and digital technologies are poised to play a major role in this transformation, empowering industry leaders to modernize their infrastructure and operational assets, improve their customer service and engagement, and better report on their sustainability progress.

The entire energy sector is at a crossroads when it comes to the adoption of AI. Energy leaders who commit now to implementing AI throughout their organization in a safe, secure, and responsible way will gain a major competitive advantage, while those that hesitate to adopt it will be surpassed in capability by their more forward-looking peers. This is especially true for generative AI: According to IBM’s 2023 CEO Study, 63% of energy industry CEOs expect to realize value from generative AI and automation in the next three years,[2]and 75% of CEOs believe that the organization with the most advanced generative AI will come out on top.[3]

Generative AI is a step change in the evolution of AI. Unlike traditional AI, which requires models to be built for each individual use case, generative AI is powered by foundation models—large neural networks trained on extensive unlabeled data and fine-tuned for a variety of tasks. These models can be applied to an array of use cases like semantic search, code generation, email routing, customer service, and improved automation for businesses everywhere. For the energy industry, it represents a tremendous opportunity to increase operational efficiency and to make strides toward the Paris Agreement’s mandate to reduce emissions by 45% by 2030.

IBM is focused on use cases that are scalable and relevant to every industry (augmenting and automating talent acquisition, customer service, and app modernization), while also developing industry specific capabilities and use-cases for generative AI. The following applications of generative AI can help organizations in the energy industry begin making gains today.

?

Putting data to action with generative AI

Energy companies can immediately embrace generative AI applications by using publicly available data for customer support and design process. This approach has the benefit of accelerating tasks without the burden of qualifying proprietary data for security, privacy, and auditability.

For example, energy companies can use this public data to:

• Enhance customer care by using existing foundational models and fine-tuning them to enable customers to search company websites and provide better insights into their accounts.
• Develop early proof of concept or minimum viable product development to avoid concerns about data security and protection.
• Utilize large language models’ (LLM) natural language processing to prepare the response to a rate case for the public utility commission.

For more advanced (and powerful) AI use, companies can apply fine-tuned foundation models to their own data sets. It’s critical, however, that organizations first establish rigorous governance and policies for the data to ensure its security and privacy. With this approach, energy companies can:

• Enable equipment operators and field service engineers to carry out complex maintenance tasks by automating and augmenting knowledge with additional instructions and manuals. Such models can help prevent plant or grid outages and streamline operational maintenance and resilience.
• Utilize image recognition for maintenance of assets – e.g. transformers, electrostatic precipitators (ESPs) that capture particles so they are not expelled into the atmosphere, etc.) and build models to monitor for cracks in facility walls and infrastructure. This could prevent an outage that costs a utility company hundreds of thousands to millions of dollars per day depending on the scale of the facility.

As energy companies further fine-tune their models, longer-term examples might include the following:

• Power and utility companies could use time series foundation models to enable better insights into long-term demand forecasting for the grid. New AI models should be able to take into consideration expected contributions from renewables based on weather and climate forecasts combined with EV adoption rates, population growth, and patterns in regulatory changes.
• All energy companies could use AI to develop project and capital program management workflows for large engineering, procurement, and construction (EPC) programs, as investing in clean energy infrastructure development increases across the entire sector and across all value chains. This is the backbone of the Energy Transition.

Getting to work with IBM’s watsonx suite

IBM’s new watsonx enterprise-ready AI and data platform is designed to help businesses capitalize on the opportunities of generative AI and foundation models and is ready for energy companies to begin transforming their operations today.

Watsonx consists of three distinct elements: watsonx.ai, watsonx.data, and watsonx.governance. Energy companies looking to increase efficiency and make strides toward clean electrification can harness these tools to improve nearly every facet of their operations.

• watsonx.ai: Energy companies can deploy this enterprise studio for AI builders to train, validate, tune, and deploy both traditional machine learning and new generative AI capabilities powered by foundation models. Organizations can set these pre-built models to work on data sets to make quicker work of a variety of tasks. For example, using public data, energy companies can speed up the onerous task of preparing a response to a rate case.
• watsonx.data: Organizations looking to optimize their operations can prepare their data for use by foundation models using watsonx.data. This repository is based on a data lakehouse architecture and open data formats and enables the creation of collated data sets for LLMs.
• watsonx.governance: Data governance is critical for the trustworthy application of artificial intelligence. With watsonx.governance, organizations can track data, curate methods and models, and enable AI that can be updated to meet evolving business regulatory requirements. This tool provides insights into how AI models are behaving, what’s driving their answers and actions, and ensures users can detect bias, drift, and other issues.

For more information on how the energy industry can take advantage of generative AI to improve their operations, see the following videos:

• Q&A with Casey Werth, GM, Energy Industry at IBM
• Q&A with Bryan Sacks, CTO, Energy Industry at IBM

[1] Our World in Data, “Sector by sector: where do global greenhouse gas emissions come from?”

[2] IBM, “CEO decision-making in the age of AI,” p.11

[3] IBM, “CEO decision-making in the age of AI,” p.6