Grid edge

AI will change the energy sector before it changes the grid

With a retiring workforce and mountains of data, next generation tools are arriving right on time.

October 26, 2023
Listen to the episode on:
 A technician monitors electricity levels in front of a giant screen.
 A technician monitors electricity levels in front of a giant screen.

After four years of developing a wholesale market trading platform using artificial intelligence, Gaiascope still watches clients struggle to trust automation over intuition. The cost is sometimes dramatic; one company lost out on $1 million after overriding recommendations from the AI.

“They probably will change the bids a few more times,” the company’s CEO and co-founder Lauren Kuntz told the audience at last week’s Transition-AI: New York conference. “I know how hard it is to let go of the reins.”

This tension — between trust and control, skepticism and optimism — set the stage for the day. Organized by Latitude Media, the event brought together utilities, startups, investors, and artificial intelligence experts across the energy and tech sector. 

Speakers described AI as an unlock, an accelerant, a tool — but many also acknowledged that adopting new, complex technologies will take time for utilities and more cautious service providers, who have good reason to value stability over speed. As conference speaker Astrid Atkinson, founder and CEO of software platform provider Camus Energy, recalled a utility executive saying to her: “I don't like to be the first or the second. I'd like to be the 23rd.” 

While utilities may be slow on the uptake, the energy sector is already changing around them. Startups and specialists in energy markets, demand response, and weather forecasting report there is plenty of data and accuracy to begin using AI as they would any other technology. And for their customers, building trust is a matter of time.

It's obvious when you stand back, said David Groarke, managing director at Indigo Advisory Group, who is partnering with Latitude Intelligence on upcoming utility AI research. “It’s a very people-focused conversation around AI.” 

The new workforce

Groarke’s research found that in the next decade, 50% of the utility workforce will retire: from engineers to operators, and field technicians to administrative staff. 

And it’s not as if utilities are currently flush with talent. In fact, they already struggle to find new recruits, with 75% reporting some difficulty and nearly 40% citing insufficient qualifications. In particular, Groarke relayed an immediate need for skills in data science, user experience design, and AI.

But this presents an enormous opportunity, he said. As a new, digitally savvy workforce enters the energy industry, they will come with a new willingness to trust AI. Of course, that doesn’t solve immediate barriers to adoption, and it will be a challenge to capture the extensive knowledge base that’s moving on. But new technologies could potentially support the inevitable skills gap.

Listen to the episode on:

The influx of expertise from the tech sector has already benefited utilities. Camus Energy’s Atkinson said that six or seven years ago, utilities were just starting to move into cloud-scale computing, and almost no one had the skills to handle big data effectively. 

“Building that familiarity and organizational fluency — with what data is, what software is, how to use it, and how to tell if it's good — is still young in the industry,” Atkinson said. “I am really excited to see more people come in from the tech side, because we're going to need it.” 

From the age of digitization to the age of automation, the convergence between the tech and energy sectors will mean fundamental shifts in what utility and energy jobs look like in-house. Do data scientists sit in digital or operations? Do they meet with customers? Have they worked at utilities before? Raiford Smith, chief utility innovation officer at AES, described his own answers to these questions, pointing to strategic shifts that need to happen at the management level. 

Though still a few years out, AI is also arriving right on time when it comes to managing the grid. As the workforce ages out, and as distributed energy resources like electric vehicles and batteries complicate day-to-day operations, AI is poised to address both problems at once.

“These kinds of next generation tools are not a nice-to-have or a novelty for how we operate our grid,” Atkinson said. “They're just on the edge of really becoming a necessity.”

Experimentation on all fronts

Running parallel to the workforce transformation is another subtler shift: the use of AI for productivity gains. 

Hanna Grene, global operations and GTM leader for energy at Microsoft, described several ways in which the energy sector can use AI for more benign tasks, including search optimization and content generation. 

In one example, Southern California Edison uploaded 22,000 regulatory documents into Microsoft’s Azure platform to create EdisonGPT, an internal tool that generates a regulatory summary and source list for policy experts, buying them back time for more creative endeavors. It’s also available externally for customers approaching utilities to receive regulatory information without waiting for their expertise.

“Utilities are still on a journey to becoming digital companies,” Grene told Latitude Media, and these applications are low-hanging fruit to start using AI tools. A spokesperson for Southern California Edison confirmed they were testing AI across the organization.

These generative AI tools are illustrative of the current moment, in which all companies embrace a period of play. They are still exploring ideas like data-sharing to build industry-wide datasets, with new founders rising up to join the ranks of the early adopters.

AI is now confidently used to maximize field worker safety and centralize information on energy markets, to design solar installation and assist utilities with electric vehicle integration. Per conversation at the conference, though founders are confident in their technology, they still need to convince each client that the automated answers are accurate and add value. In some cases, such as forecasting, seeing the results helps build trust over time, Kuntz said during a panel on energy markets. 

Nancy Hersh, founder of Applied Data Science and a conference attendee, said she believes that while this period of excitement is good, more companies need to focus on publishing measurements of how effective their AI really is in order to move forward efficiently. 

“Once you let a thousand flowers bloom, if you can figure out which of the thousand are best, put more resources towards those,” Hersh said. Companies can then use the resulting data “to make each of the flowers a better flower,” she added. 

Despite the potential, though, AI enthusiasts and practitioners at Transition-AI: New York are still largely an exception to industry norms. 

Research from Indigo Advisory found that at least one-fifth of decision makers in the utility sector are still cautiously observing developments in AI, without taking action of their own. Meanwhile, roughly one-third has moved into strategic planning for the technology, and another third is dabbling in it, without substantial investments.

While industry stalwarts like Microsoft and GE have been employing some form of AI for decades, utilities are only now getting comfortable with the technology. As gatekeepers of decarbonization in the power sector, though, that utility adoption of AI is critical for maximizing the value and reliability of renewables and storage. 

Smith of AES had a few words for his peers about the transformation: “Don't be afraid of the change.”

Editor's note: This story was updated on October 26 to make two clarifications: first, that Astrid Atkinson was the source of the quote in the fourth paragraph, and also that insights on founders' need to convince clients of the usefulness of AI originated from conversations at the conference, as well as from Lauren Kuntz' comments on a panel.

No items found.
No items found.
No items found.
No items found.
big data
demand response
energy distribution
smart grid