Utilities Embrace Copiloting with Artificial Intelligence, Parachute-Free

Teri Viswanath

April 25, 2024

Generative artificial intelligence is having a “throw everything at the wall” moment, with the market’s current enthusiasm similar to the halcyon days of the dotcom era. While 61% of CEOs polled by PwC are counting on generative AI to measurably boost the quality of their products and services, utilities are a particular standout.

Nearly three-quarters of energy and utility companies have implemented or are exploring using AI in their operations, according to a study by IBM. From DER integration and EV charging to power line inspections and wildfire detection, utility industry leaders are rapidly evaluating AI as a possible solution for a veritable smorgasbord of grid challenges.

However, generative AI can only address the finite problems it has been specifically programmed and trained for. It relies heavily on clean and relevant data, with the kryptonite-like effect of bad or false data creating AI hallucinations. Still, the message to “curb your enthusiasm” hasn’t swayed investment. Bloomberg Intelligence predicts that “generative AI is poised to expand its impact from less than 1% of total IT hardware, software services, ad spending, and gaming market spending (in 2023) to 10% by 2032.” Bloomberg sees this year as the steepest slope in the decade-long, open-checkbook spending spree. That said, the practicalities of securing the next tranche of hefty spending commitments, at least for utilities, will presumably be linked to more down to earth measurable results. And given the rate of current spending in this space, reconciliation is likely coming sooner rather than later.

Sharp thought leaders expect 2024 to be the utility year of AI reckoning. On our CoBank Power Plays podcast, Watt's Up with AI? The Role of Artificial Intelligence in the Future Grid, Dr. Jeremy Renshaw of the Electric Power Research Institute stressed the importance of understanding the types of problems best suited for generative AI or the most efficient tool to apply — and, conversely, when it makes sense to look elsewhere. Renshaw views AI as not a ready-made solution, but a versatile tool that requires an informed user. “We've often seen that training our existing workforce to understand how and when to use AI is very valuable, so they'll understand what the answer should be or what the range of possibilities should be so if they're getting a bad answer, they'll be able to identify that easier,” he said.

The electricity sector is pre-wired for generative AI

With this in mind, we posed the simple prompt to ChatGPT, Microsoft Copilot and Google Gemini — “Help me write a 200-word outline on how artificial intelligence and machine learning can be applied to the power utility space.” These are three of the most popular large language models (LLM), a type of generative AI specifically designed for tasks revolving around natural language generation and comprehension. Admittedly, this is a biased test, as operational segments across the utility value chain are uniquely well defined and documented compared to other industries — which likely explains why utilities are test-driving the technology at a faster rate than their peers. Whether it is generation, transmission or distribution, the pre-defined electricity use-cases make the industry ripe or at least pre-wired for applying generative AI. But to Renshaw’s point, the key for unlocking value from the technology is having a knowledgeable utility expert to identify where the AI response is possibly incomplete or, worse yet, incorrect (an AI hallucination).

So, let’s put Renshaw’s observation to the test: Review the three AI generated responses below and evaluate whether generative AI — at least LLM — will be a positive, disruptive force for utilities. Do the AI-generated answers feel complete and correct? Can you spot the distinction between these platforms? Are the responses differentiated enough to tell? And, most importantly, would these applications provide some work efficiency or support your daily business tasks that require creativity, thought organization, research, writing and editing?

Since ChatGPT was released in November 2022, the field of generative AI technology has rapidly advanced, with greater LLM competition, wider adoption and the launch of a diverse array of non-LLM mainstream applications. Taking stock of these developments, Flex.OS recently reported on the most used AI tools by analyzing web traffic and search rankings. They noted that current use of AI tools is concentrated around just a few applications, with ChatGPT monopolizing over 50% of would-be AI-user’s monthly visits. This concentration climbs closer to 70% when the other top-five general GPTs — such as Microsoft Copilot and Google Gemini — are included in their count. Breaking AI site visits down by age group, we see that the youngest AI users spend a significant chunk of time with chat buddies and homework help, which might suggest a generational shift in the growing dependence on this technology.

Generative AI success linked to labor management (yes, another LLM)

With an increasing share of Americans utilizing these tools, there is little doubt that generative AI will impact our future workplace, but in what ways? For the next generation of workers, as the Flex.OS authors summarize, growing up with an AI friend and “getting highly personalized help when you're dealing with a school assignment warms you up for AI to be alongside you in the workplace.” Makes sense, but exactly how will generative AI sit shotgun or copilot us into our future? Fortunately, given both the heightened interest and investment in generative AI, early studies have already produced a growing body of research on labor productivity.

One particularly thoughtful study by the National Bureau of Economic Research (NBER) measured productivity gains in a customer service department of a Fortune 500 software firm after the introduction of a generative AI support tool. The company pre-trained the open AI LLM on data from 5,000 customer service agents and specifically designed it to lend support to existing agents dealing with in-bound customer inquiries. The results were impressive, with an average 14% uptick in the number of hourly customer resolutions for the customer service department and other time efficiencies. Employee satisfaction also rose with the use of these resources. Not surprisingly, the greatest benefits occurred with less-skilled and less-experienced staff with a 34% increase in the number of issues they were able to resolve per hour.

Based on this study, we are reasonably confident that the utility generative AI customer service programs will witness some level of success this year. And depending on the level of that success, utilities could greenlight follow-on investment for 2025 and beyond. But the NBER researchers' public service message is that the success of the software company program was strongly tied to the contributions (AI training data) of the more experienced, skilled customer service agents. Their results specifically raise concerns on the need for generative AI program design to properly incentivize workers to contribute and readily sign up for this copiloting future. And this is where it gets tricky.

The NBER authors note, “we hypothesize that part of the effect we document is driven by the AI system’s ability to embody the best practices of high-skill workers…and make it available to other workers.” Not to throw shade, but it feels like we might be collectively experiencing another Napster moment with these generative AI copilots — where smart, creative people become very unhappy about their “GPS skills” being used to pilot GPT.

The other labor-related problem is figuring out the most effective application for generative AI within utilities. In a blog dedicated to figuring out how AI will impact work, education and life (On holding back the strange AI tide), Dr. Ethan Mollick, a professor at The Wharton School of the University of Pennsylvania, identifies three common pitfalls for corporate behavior associated with generative AI:

1. Companies will choose to IGNORE the technology. Bad idea. As Mollick points out, “Ignoring AI doesn’t make it go away. Instead, individual employees will find ways to use AI to enhance their own jobs. They won’t tell the organization’s leaders about what they are doing, because they worry about being punished, or that others will value their work less.” If companies are unaware of how AI enhances worker output, they cannot leverage these learnings.
2. Companies will choose to BAN the technology. Once again, a misstep. Mollick writes, “This is usually in response to well-intentioned, but sometimes technically incorrect, legal opinions. When companies ban AI, workers continue to use it on their phones and home computers. And they still don’t tell you what they are doing.”
3. Companies will choose to CENTRALIZE the technology. Good intentions, but bad execution. According to Mollick, “Centralization is what organizations are used to doing when faced with a new technology. By trying to make AI like all other technologies, companies are ignoring how transformative it is. Only innovation driven by workers can actually radically transform work, because only workers can experiment enough on their own tasks to learn how to use AI in transformative ways.”

Mollick’s last bit of wisdom leaves us off where we started, pointing out just how important it is to have worker buy-in for the success these of workplace generative AI initiatives. For rural electric cooperatives, there can be tremendous upside for striking the right balance. NRECA data says that 19% of electric cooperative workers are retirement-eligible in the next five years, compared to 9% across the entire energy industry. Consequently, investigating pathways to continue to secure the best talent and create new labor efficiencies will continue to be a pressing priority. But here too, we are optimistic.

Co-ops have traditionally been ahead of the curve with applying new technologies, especially when confronted with reliability or affordability opportunities. Case in point: Co-ops were early adopters of advanced metering capability to reduce the costly labor requirements of on-site meter reading. Thereafter, co-ops embraced AMR and then AMI as an efficient approach to resource scarcity. Consequently, generative AI might be the next technology bridge to traverse on the road to keeping member rates affordable. And there are so many well-defined areas where systems can achieve efficiency. On our podcast, Utilidata CTO Marissa Hummon spoke of potential utility efficiency gains, of layering on communication and computation technologies, with the endgame of achieving “double the amount that the distribution grid can handle without building out new lines.” Woah.

As our Microsoft Copilot sums it up, AI applications hold the promise of empowering power utilities to operate efficiently, reduce downtime and enhance overall grid performance. Or, as we like to think of it, a future where we can confidently co-pilot with AI, without the need for a parachute.