AI is coming to make our lives better!

AI is coming to destroy us all!

Ever since ChatGPT, the artificial-intelligence-powered chatbot, took the internet by storm in late 2022, stories about the amazing possibilities and terrifying pitfalls of AI have dominated conversations in the media, at the workplace and around the dinner table.

But this furor obscures the fact that in the energy industry, AI is already here, in ways big and small.

And for electric cooperatives, its importance is almost certain to grow.

“It’s going to be transformative,” says Venkat Banunarayanan, NRECA’s vice president for integrated grid.

Banunarayanan and other experts note multiple ways in which AI will impact cooperatives and other electric utilities, while cautioning that its growth does come with challenges and potential risks.

The power industry is actively exploring AI’s potential, according to a survey by Capgemini Research Institute, an online think tank. The survey found that a third of utilities and energy companies worldwide are doing pilot projects on “generative AI,” a form of AI that can generate its own text, images or other content. It’s the kind of AI that gives ChatGPT its ability to interact with the public. The survey also found that 70% of organizations in the energy and utility sector say the benefits of generative AI outweigh the risks.

Here are some key ways industry experts expect AI to impact utilities:

Grid management

“The grid has already been described as the most complicated machine in the world,” says Jeremy Renshaw, senior technical executive for the Electric Power Research Institute (EPRI). “The grid of the future is going to be significantly more complicated.”

The rapid growth of renewables and other sources of distributed generation, along with advances in demand control and shifting load profiles tied to electric vehicles, battery storage and other changes, all have made managing the power grid more complex.

The scale of that complexity and the speed at which adjustments will need to be made would require a highly trained workforce far larger than the industry can afford, says Renshaw, who spoke at an online press briefing earlier this year hosted by the United States Energy Association (USEA).

But with the help of AI, the potential benefits of a grid that integrates a large amount of distributed generation and demand response can be more fully realized.

“If we can utilize these technologies properly,” Renshaw says, “it can really help us manage a very complicated grid to provide low-cost, decarbonized energy to everyone.”

But even as AI assumes a larger role in managing the grid, John Savage, a professor emeritus of computer science at Brown University, believes it’s always going to be necessary to have a human in the decision-making loop.

“We have a complex and sensitive system that should not be fully entrusted to an AI system,” he says.

Savage, who also spoke at the USEA briefing, cautions that AI systems have reached such a level of complexity that even their creators “do not understand why their systems behave—and misbehave—the way they do. As a consequence, my view is that one has to be very careful in deploying systems using AI.”

On a smaller scale, however, AI is already being baked into utility and consumer technology. One example is the Sense Home Energy Monitor, which uses machine learning to determine the power use of different home appliances, providing consumers a much more granular look at how they’re using electricity.

More lies ahead, says Tolu Omotoso, NRECA’s director of energy solutions. “The next generation of meters are going to have all these integrated, AI-enabled capabilities to help people personally manage their energy use.”

Member service

Banunarayanan points to two ways AI can be key to improving member services.

“One is understanding your members better,” he says. “Going deeper into what their needs are, what their hopes are, based on AI analysis of all the different kinds of interactions you have with each consumer.”

Anyone who has shopped for a product online only to be inundated with email offers for similar products knows machine learning already plays a role in targeting consumers. But AI’s ability to learn from data gathered across a variety of member-contact points could provide electric co-ops a greater opportunity to respond to their individual needs.

The second way AI can aid member services, Banunarayanan says, is giving the co-op a better idea of what initiatives are likely to be successful.

“If you are designing a program, an EV charging program or a rate incentive program, for example, you can use AI to predict the success of those programs,” he says.

The conversational ability of generative AI chatbots such as ChatGPT also hold the possibility of transforming member services. NRECA’s Business and Technology Strategies department, for instance, had summer interns build a co-op-focused chatbot pilot to explore its potential for member co-ops.

“We’re seeing more and more AI-assisted help, especially in areas where there’s high turnover like call centers,” says Marc Spieler, global business development director for Nividia, a leading AI company. “We can use AI to look at data and make recommendations to the people on the [business] end of the phone on how to provide a better customer experience.”

Weather forecasting

AI’s ability to learn from large data sets and refine its analysis over time as more information comes in makes it a valuable tool for weather forecasting. AI has been used in forecasting for some time, but the latest advances in artificial intelligence promise further advances in the scale and accuracy of these forecasts.

Several technology firms, including Google and Nividia, are using AI to build better forecasting models. Nividia, for example, is in the process of building “Earth 2,” a digital twin of the earth that will be used to model the longer-term impacts of climate change.

Such simulations could also help electric co-ops and other utilities determine where it makes the most sense to make future investments in grid resilience and solar and wind generation.

In 2022, Horry Electric Cooperative, based in Conway, S.C., conducted a pilot project using a combination of AI and satellite imagery on a feeder line for right-of-way inspection.

The co-op decided to stick with its established vegetation management program, but saw value in the approach, in which AI drew on the images and historic weather and outage data to predict the risk of trees or limbs coming down.

“The technology is very promising,” says Kevin Jordan, Horry Electric supervisory engineer, “particularly for cooperatives trying to decide what areas to target first.”

Image recognition that equals or surpasses human capability is a challenge for AI “because how our eyes and our brain interpret images is very, very tricky for AI to emulate,” notes Banunarayanan. But as the capability grows, he says, it could relieve co-op personnel of much of the time-consuming work of reviewing drone images or other data as part of regular inspections.

Challenges

With all of AI’s potential, assuring the quality of the data it consumes will be paramount, says Carter Manucy, NRECA senior manager for cybersecurity.

“If we are getting a lot of data coming in and someone is able to get into it and corrupt it, the question is, ‘Are we getting bad signals because of bad data?’” he says. “Data manipulation is a big concern.”

Data security also becomes critical. Right now, Manucy notes, none of the information the public is sharing with ChatGPT is “controlled.”

“You don’t know where it’s going,” he says.

As electric co-ops turn to similar chatbots to improve member services, he points out that assuring the security of sensitive consumer data will be a priority.

Co-ops need to be aware of AI’s challenges and should develop plans to meet those challenges, Manucy says. But avoiding the technology or trying to prevent employees from using it is unlikely to be a feasible long-term approach.

“Just like the cloud,” Manucy says. “AI is here to stay.”