By most assumptions, a data scientist with a PhD in AI and machine learning would be the biggest cheerleader for Generative AI. Meanwhile, a change management expert would naturally push back on rapid AI adoption.
But when Brittany Stone, lead for change management at Method, and Gareth Jones, data and AI solutions lead, sit down to discuss the topic, there is a strong degree of alignment: Companies are over-indexing on GenAI when traditional machine learning and automation often deliver better results.
No one is disputing that AI will change how businesses operate. The surge in GenAI has pushed executives to finally treat data as an asset rather than a liability. But the fixation on GenAI, and particularly Large Language Models, has created blind spots, shutting out value-driving opportunities that simpler approaches can deliver faster and safer.
The GenAI Hype Cycle
GenAI and LLMs come with a high bar. They need strong infrastructure and massive amounts of high-quality data, and they raise privacy and regulatory concerns. And despite the buzz, we still lack proof of long-term business impact due to their recency.
Looking at the other side of the value/effort exchange, ROI opportunities are narrow when we look solely at GenAI. In some more digitally native industries like EdTech, we see broader systemic-level opportunities to leverage GenAI; however, in physical industries like rail or energy, we are often stuck focusing on use cases at an individual contributor level.
This, fundamentally, lowers the glass ceiling for overall impact. Ninety percent time savings on a two-day task with a 10-person team doesn’t compare to 30% on a two-month task with a 100-person team.
On top of this, we must consider the change management perspective; a solution will only ever drive value if people use it. And, no surprise, we see limitations here as well with GenAI.
GenAI adoption is polarizing. It generates strong opinions, reluctance, and caution in ways traditional automation doesn’t.
When people ask for GenAI solutions, they often don’t know what they’re actually asking for. They’ve heard the hype and feel the pressure to keep up. “Everyone else is doing it! If we don’t, we’re behind!” Companies have developed GenAI Myopia.
This FOMO drives too many strategic decisions. If your vendors and partners are not challenging you to explain why you are pursuing GenAI capabilities, they are doing you a disservice.
Now, let’s compare that to traditional machine learning and automation. The bar to entry is lower. You can meet clients where they are. Development is faster, risk is lower, and there are no hallucinations to manage. Business impact is broader, with higher potential ROI. And from a change management perspective, you’re improving existing workflows rather than forcing people to rethink how they work.
Energy’s Double Burden
Now, let’s take the energy industry as an illustrative example.
The energy industry is particularly interesting as it faces a unique double burden. The industry is undergoing an enormous surge in demand because of AI. As AI adoption skyrockets across industries, the demand on the grid has become exponential.
The infrastructure challenge is staggering. Everything an energy provider did, they now have to do three to five times over in a fraction of the usual timeframe. The same number of substations, poles, and wires they’ve installed over the last few decades now needs to be installed in just a few years.
This creates a feedback loop: AI drives demand on the grid, which drives the need for AI to manage that complexity. Energy companies are simultaneously feeding the AI boom and grappling with the challenges of using it to keep up.
Despite being perceived as technology laggards, energy companies were early adopters of AI because of the grid’s complexity. When you’re managing what may be the physically largest and most complex machine humans have ever built, traditional approaches hit their limits. AI performs rapid computations and analyzes specific data points to help manage grid load and predict maintenance needs.
But, as with many other industries, the conversation has shifted from “How do we use AI to solve our complexity problems?” to “How do we implement GenAI?” That shift narrows the focus and misses practical opportunities.
When Traditional ML Wins
Method recently worked with a regional transmission authority on exactly this type of challenge. When someone wants to connect new generation to the grid, they submit an application. The approval process takes 15 to 16 months, with three to four months spent just writing reports.
At first glance, report writing seems perfect for GenAI. After all, we are generating content. It’s in the name, right? But actually, only a small amount of the total time is going toward generating content. Most are lost to extracting and aggregating data from simulation outputs and reports. Besides, the reports are standardized and follow a set structure.
The work is documentation, not invention. So, does GenAI really make sense?
Suppose we take a more traditional AI and automation solution instead. Traditional AI can extract data from reports. Simple data engineering can structure this extracted data. Automation can be used to run standard analyses and populate templates to generate reports.
What took three months now takes one day. Humans review for accuracy, and because the output follows templates, that oversight is straightforward.
One developer can build this solution in about 10 weeks and knock off two and a half months from a 12-week process. That’s measurable impact with predictable timelines and clear ROI.
If you tried to solve this with GenAI, the timeline gets murky. The technical requirements jump. The confidence in the outcome drops.
Now, let’s look at all this technology from a humanistic perspective.
The Change Management Challenge
A recent study from METR examined how early-2025 AI tools affect experienced open-source developers. The results were surprising: When developers used AI tools, they took 19% longer to complete tasks. Even more striking, developers believed AI sped them up by 20%.
The gap between perception and reality uncovers a critical change management issue.
From a change management standpoint, AI creates concerns on both ends of the adoption spectrum. Early adopters risk over-reliance. In expertise-driven industries like energy, that expertise matters. You need people to show up to work with both an LLM and a brain.
People need a basic understanding of what an LLM is, what it isn’t, and how it works. When you understand the technology at even a conceptual level, it feels less like magic. You know what it’s good at, what it’s not good at, how to get the most out of it, and where you need to fill in the gaps.
On the other end, skeptics may shut down entirely or even sabotage implementations. There are reports of experts purposely training models poorly because they fear replacement. The response has to be “show me, don’t tell me.” Building trust takes time and demonstrated value.
Traditional automation faces less resistance. When you automate document gathering so people can focus on analysis, you’re not asking them to change how they think. You’re removing the hunting so they can do the thinking.
The pattern we illustrated here repeats across manufacturing, logistics, and finance.
The energy sector’s aging workforce presents another consideration. The industry faces a talent shortage as experienced workers retire and fewer young people enter the field. Some companies have explored using LLMs to capture decades of institutional knowledge.
But the practical challenges are steep. Getting tenured employees to document their expertise into an LLM, parsing through that information, and teaching people to incorporate it into their workflows requires significant change management. The juice may not be worth the squeeze, at least not yet.
Moving Beyond the Hype
Declaring yourself “pro-GenAI” or “anti-GenAI” misses the point. AI is here. It’s good at specific things. The question is how to deploy it strategically.
The goal isn’t to abandon GenAI. There are legitimate use cases where it provides value. But companies need to approach AI strategically, matching the solution to the actual problem rather than forcing GenAI into every conversation.
Ask what you’re trying to accomplish. Then determine if GenAI, traditional ML, or basic automation delivers the best ROI. Often, simpler approaches deliver more impact faster and with less risk.
The companies that succeed will look beyond the hype cycle and focus on actual business value. They’ll use GenAI where it makes sense and lean on traditional ML and automation where those approaches deliver better results.
The conversation needs to shift from “How do we implement GenAI?” to “What problem are we solving, and what’s the best tool for that job?”
When you’re managing critical infrastructure with life-or-death consequences — or leading a successful business in any industry — tuning what already exists often beats a complete overhaul. Method works with energy companies and other complex enterprises to identify the right AI approach for specific challenges.
If you’re looking for a vendor who will have candid conversations about GenAI and cut through the hype cycle, let’s talk about what actually works.