AI, Utilities, and Solving Key Business Problems: Part 1 – AI Isn’t a Magic Wand 

Every few years, a technology gets so overhyped that just buying it can seem like a valid business strategy. And right now, that technology is Artificial Intelligence (AI).

After 20 years of helping large organizations navigate technology transformation, we watched a troubling pattern play out with many transformational initiatives, including enterprise resource planning (ERP), cloud, and now AI. It goes something like this: a vendor demo is impressive, a budget gets approved, and 18 months later, stakeholders (often on the front line) are asking why nothing seems to have changed. 

At UDC, we’ve been engaged in the wake of these initiatives and have found a common thread.

The problem isn’t the technology; it’s how organizations think about it.

What actually is AI??

If you believe in magic, you might consider AI as the magic wand. Wave it at a problem, and it will go, poof! Let’s admit that AI is not magic. 

So, what is it, what’s under the hood? If you can get past the hype, AI is, at its core, pattern recognition at scale^[1]. Feed it enough data, define what you want it to find, and it gets remarkably good at spotting things we humans would miss or couldn’t process fast enough to act on.

That’s genuinely powerful. But it’s also specific. While AI models are changing (and fast), the baseline remains the same. AI doesn’t reason. It doesn’t understand your business. Like any new employee, it doesn’t necessarily know what “good” looks like unless you tell it — repeatedly, with data, over time. 

When you think about AI from this lens, it becomes a different conversation.

The AI question – How you ask it matters

Right now, many organizations are asking:

“How can we use AI?” instead of starting with, “What problem do we need to solve?”

This has been a common mistake for years and, at the current scale and pace, can be more costly than ever.

When the technology leads the strategy, you end up with features that don’t actually help to get results. Dashboards that surface insights nobody acts on. Automation that speeds up a broken process instead of defining a root cause and fixing it. The algorithm is working; however, the business isn’t moving.

Research confirms this pattern: Jon Garcia, in his McKinsey interview^[2], reports that 70% of digital transformations fail to meet their objectives, while Bain & Company’s 2024 study^[3] found 88% of business transformations fail to achieve their original ambitions. The primary culprit? Something UDC has understood for years- it’s not the technology itself, but how organizations approach implementation.

The Right AI questions to ask

Instead, start with: 

“What decision do we need to make better, faster, or at lower cost?“

Then ask: 

“Do we have the right data to support it?” and “Will our operating model allow us to act on the answer?“

If the answer to both is yes, then AI is a great option for acceleration. If you answer no to either of these questions, AI is not going to be able to help solve your decision problem.  

What this means practically

Real value from AI right now isn’t materializing at companies that deployed it the fastest. Instead, it’s organizations that:

• Identified a specific, high-value decision that was being made slowly or inconsistently
• Had (or built) the data infrastructure to feed it 
• Are willing to change how people and processes work to support the effort

That last part is where most technology transformations fail – not just AI. Boston Consulting Group research^[4] found that roughly 70% of AI implementation challenges are related to people and processes, not technical issues. Organizations that invest in change management are 1.6 times as likely to report that AI initiatives exceed expectations.

You can build the most advanced solution possible, but without the proper organization in place to implement it, change will not happen. 

AI is a powerful tool. But a tool doesn’t transform an organization. The decisions your people make with it do. 

Learn more about the use and exploration of AI at UDC; view a copy of our AI/ML policy and watch our presentation Harnessing AI for Utility Network Transformation Success.

Author’s Note: I worked with AI to write this article—ideas and content provided by me, polished by Claude.

Next in this series: Part 2 – Utilities Have Been Doing AI for Years. Utilities have more data than almost any other industry, and yet so much of it goes unleveraged. In our next post from TJ Houle, discover how organizations can take advantage of the untapped potential.