The “three Rs” of artificial intelligence – relevant, reliable, responsible – are as fundamental to a utility’s successful implementation of AI as the original three Rs are to successful human education. “Relevant” is the most important of these Rs at this early stage in the evolution of generative AI in the utility industry.
Relevance hinges on a utility’s needs, the investments involved, and the existence of AI-driven solutions to meet those needs in ways that boost the business, improve service, smooth operations, cut costs, or, ideally, all of the above.
At the moment, few AI solutions can boast such outcomes. But they’re coming in a matter of months, and it looks like customer service functions will be the early winners.
A large language model may answer all queries with idiomatic precision, but the value of those answers will depend on the agent’s access to accurate customer and business data, and the more of it, the better:
Note that the “collecting” of data doesn’t mean it must all reside centrally. Federated learning, a machine-learning approach that lets AI models be trained collaboratively across decentralized data sources, means utilities can build centralized AI models that comply with data privacy and data residency laws while leaving the underlying business data in place.
In order to get the biggest business benefits from AI in customer service, businesses need to look at the entire service value chain and figure out how to use AI across the various support stages.
Now, let’s talk AI in utility customer service. Self-service AI assistants for utilities bear little resemblance to the largely ineffective chatbots of the past, and unfortunately today. Tomorrow’s versions will bolster utility customer information and service solutions in two key ways.
Second, self-service AI assistants for utilities do more than just ask questions. They keep tabs on customers and reach out proactively.
For example, if usage spikes beyond historical patterns, the system considers possibilities such as infrastructure issues or high temperatures driving more air conditioning use. If those don’t explain the greater usage, the system will reach out to the customer to note the uptick, inquire about the causes, and, if relevant, suggest alternatives.
If, say, the higher usage came from a new electric vehicle being plugged in, the AI assistant could propose solutions such as alternative charging times to reduce costs, or installation of solar panels to increase energy efficiency. All that stands to improve customer loyalty and boost revenues while reducing customer service costs by an estimated two-thirds. And it will happen without the intervention of a human service agent.
A related, emerging gen AI application for utilities is intelligent dispute management. This uses sentiment analysis to prioritize incoming requests, summarize inbound emails, cut down on the manual effort needed to track down invoice data, reasons for the dispute, route questions to the right teams, and formulate outbound responses. This will save precious minutes per dispute case, increase accuracy, and speed resolution.
Gen AI will only be part of the story. Utilities have been using machine learning and computational AI for years in predictive maintenance and other operations. These forms of AI, in concert with gen AI, will be crucial to what will be most impactful future AI application for utilities: the holistic coordination and management of distributed energy resources.
These comprehensive energy management systems will be able to track thousands of energy sources and uses. They’ll establish the basis for nimble capacity management and dynamic pricing models that adjust energy prices (and, therefore, steer production as well as usage patterns) based on market conditions, consumer demand, weather patterns, and other inputs.
They’ll be indispensable to enabling the energy transition and meeting ambitious carbon-reduction targets, so it’s no surprise that a host of global IT giants are working with utilities to spec out and design them. When they’ll roll out is still an open question. But given the profound need, the enormous potential market, and the existence of enabling AI technologies, bet on sooner than later.
The utilities landscape is changing fast as more consumers, both commercial and residential, are starting to capture and store their own energy.
The critical nature and massive economic impact of utilities make reliable, responsible AI an absolute priority. But for now, utilities are rightly focused on relevance, and that’s the way it should be.
Editor’s Note: This article was first published in POWERGRID and is republished here with permission.
