How AI is changing utility regulation

I have been thinking about several issues we should be discussing now, while we are all still figuring out how AI will fit into utility regulation. 

What happens when people stop coming to us for information? 

Utilities and regulatory agencies spend considerable time making information available to the public. We maintain websites, publish reports and notices, answer questions, and try to explain what we are doing and why. We do all of this with the expectation that when people want to know something about us, they will eventually come to one of those sources. 

But increasingly, they may not. AI may be talking about your organization before you get the chance. 

Suppose someone wants to know about the major utility issues in Jamaica. Instead of going to the regulator’s website, that person might simply ask Google or ChatGPT. Within seconds, AI will provide an answer about outages, electricity theft, infrastructure, scarcity, distribution problems, and whatever else the AI predicts. 

Maybe the answer is excellent. Maybe it isn’t. Either way, the person asking the question may feel that he or she has an answer and moves on without ever seeing what the regulator itself has to say. 

AI is becoming your spokesperson. I wonder how many utilities and regulators are thinking about this yet. 

For years, organizations have worried about what appears on their websites and how they show up in search results. We may now need to worry about something different. What does AI understand about our organization, and what will it say about us when someone asks? 

This seems particularly important for regulators because communication is already difficult. Regulatory issues can be technical and complicated, and different stakeholders often see the same issue very differently. Now we have an additional participant in that process — AI — interpreting and summarizing information before it reaches the person who asked the question.  

Where do we actually want to use AI? 

I don’t think the answer is to become afraid of AI. There are many things I would be very happy to have AI help us do. 

Anyone who works in regulation knows how much time can disappear into research, data analysis, benchmarking, reviewing documents, and repetitive administrative work. If AI can help people do some of that work faster and better, wonderful. It may give us more time to work on the things that require our attention. 

I am less convinced that we should automatically use AI for the parts of our work that involve relationships. Maybe we don’t want AI representing our organization. 

Most of us have probably had the experience of calling customer service and finding ourselves talking to an automated system when what we really wanted was a person. Sometimes AI is perfectly fine. I don’t need a human being to tell me my account balance. But my feelings change pretty quickly when I have an unusual problem, and the system doesn’t understand what I am trying to explain. 

Imagine that experience when your electricity is out. Or when you are trying to understand a regulatory decision that affects your business. At some point, people want to know that there is another person listening, exercising judgment, and taking responsibility. 

That is particularly important in utility regulation. Much of what regulators accomplish depends upon relationships, trust, judgement, legitimacy and accountability. Regulators need operators to share information with them. They need to understand the concerns of consumers and other stakeholders. They also work in environments where political pressures can be significant and where decisions have real consequences for people. 

Perhaps one of the best things AI can do for us is free people to spend more time on those human parts of their jobs. 

AI adoption is also a leadership challenge 

There is another issue that I think deserves more attention. We sometimes talk about AI adoption as though the important question is whether an organization has the technology. I suspect the more difficult question is whether the organization knows what to do with it. 

Someone has to decide which tasks should change. Someone has to decide how much risk is acceptable, when an AI result needs to be checked, and which decisions still require human judgment. Someone also has to help employees learn new ways of working without losing important knowledge and experience. 

Those sound like leadership responsibilities to me.  

For utilities and regulators, the stakes can also be unusually high. We cannot assume that experimentation that works for a retailer, marketing company, or social-media platform will work equally well for an organization responsible for electricity, water, telecommunications, or regulatory decisions. 

So I would not worry too much about whether your organization is the first to adopt every new AI tool. I would worry about whether your people are learning. Are they experimenting where it is reasonable to experiment? Are they thinking critically about the answers AI gives them? Do they know when to question those answers? Are managers creating an environment where people can learn how to use these tools responsibly? 

And perhaps most importantly, are we using AI in ways that give people more time to do the work that people are particularly good at doing? 

These are questions we will be exploring more at PURC. I don’t think any of us has all the answers yet, and given how quickly the technology is changing, I would be suspicious of anyone who says they do. 

But I do think we should be asking the questions. 

AI will become part of the everyday work of utilities and regulatory agencies. How useful it becomes will depend largely on the people using it and the choices they make along the way. 

We often hear that “we are in the age of AI.” At PURC, we think that “we are in the age of people using AI.”