The Dormant AI Capability Hiding in Plain Sight

Most organizations have an AI adoption story. Almost none have an AI acceleration story. The difference between those two sentences is where competitive advantage is being won and lost right now.

Last month, OpenAI and Harvard economist David Deming released the largest study ever conducted on how people actually use ChatGPT. The researchers analyzed 1.5 million conversations across three years of widespread AI availability. The finding was striking in its simplicity: nearly 80 percent of all usage falls into three categories — guidance, information seeking, and writing. After three years, after billions of dollars in enterprise investment, after thousands of hours of AI training programs, most people are using the world's most capable AI system as a slightly smarter search engine.

Read that again. Not as a technology failure. As a leadership failure.

The capability is there. The utilization isn't. And the gap between those two realities is not a technology problem — it is a human transition problem. More specifically, it is a leadership problem about who gets empowered to close it.

ADOPTION IS NOT ACCELERATION

Most organizations have confused these two things, and the confusion is expensive.

Adoption is measurable. Licenses purchased. Training sessions completed. Usage rates tracked. Adoption looks like progress on a dashboard. It feels like momentum in a board presentation. And for the vast majority of organizations, it is where the AI story ends.

Acceleration is something different. Acceleration is what happens when AI stops being a tool workers use and starts being a system through which workers build — new capabilities, new processes, new competitive positions that did not exist before. Adoption gets you to a smarter search engine. Acceleration gets you to a structurally different organization.

Adoption gets you to a smarter search engine. Acceleration gets you to a structurally different organization.

McKinsey's 2025 State of AI research is precise about what separates these two outcomes. Only 21 percent of organizations using generative AI have fundamentally redesigned their workflows around it. MIT's NANDA initiative studied more than 300 enterprise deployments and found that 95 percent produced zero measurable P&L impact. Zero. Not minimal. Not disappointing. Zero. Organizations are spending real money on adoption and capturing almost none of it as enterprise value.

This is the AI utilization gap. And it is not closing on its own.

THE VALUE HAS MOVED — MOST LEADERS HAVEN'T NOTICED

For decades, technology advantage lived with the technologist. The people who could build and deploy systems held the leverage. They sat closest to the capability. Everyone else waited for what the technology teams decided to build.

AI has inverted that equation completely. And the inversion is moving faster than most leadership teams have registered.

The value in an AI-enabled organization no longer lives in the technology. It lives in the domain expertise of the people closest to the work. The operations manager who knows exactly where the process breaks down in ways no system has ever captured. The sales leader who understands what the CRM will never tell you about why a deal actually closed. The service professional who hears what customers are really asking for three questions before they ask it. The analyst who has spent fifteen years developing intuition about which numbers actually matter.

These people have always held the most valuable knowledge in the organization. What they have never had, until now, is a tool capable of matching the speed and complexity of what they know.

The value is shifting from the technologist to the domain expert who can apply AI to accelerate innovation.

That shift is not a prediction. It is already underway. The question is whether your organization is positioned to capture it — or whether that value stays dormant while your competitors figure it out first.

THE COMPRESSION NO ONE IS TALKING ABOUT

Here is the insight that most AI conversations miss entirely, because most AI conversations are still about efficiency.

AI has not just made workers faster at existing tasks. It has compressed the time and capital required to build entirely new capabilities.

What used to require a development team, a six-month roadmap, a technology budget, and an IT project plan can now be prototyped by the person closest to the problem in an afternoon. A supply chain manager can build a decision support tool for their team without writing a line of code. A regional sales leader can create a coaching system that encodes the judgment of their best performers. A finance director can build a scenario modeling workflow that used to require a consultant engagement.

This is not productivity. This is capability creation at a speed and cost that has never existed before. And the people who can unlock it are not in the IT department. They are in every department — sitting on decades of domain expertise, operational intuition, and process knowledge that AI can finally access and amplify.

BCG's research on what they call 'future-built' organizations — the five percent generating 1.7 times the revenue growth and 3.6 times the shareholder return of their peers — confirms the pattern. These organizations are not simply using more AI. They are redesigning how work gets done, built around what their workers actually know. The technology is the accelerant. The domain expertise is the fuel. You cannot accelerate without both.

THE LEADERSHIP PROBLEM IS NOT AT THE TOP

Wharton's research on enterprise AI adoption named something important that most organizational AI diagnostics miss. They identified what they called a 'donut hole' in AI adoption — C-suites investing heavily at the top, younger workers using AI natively at the bottom, and middle management — the people responsible for orchestrating actual work transformation — stalled in the middle.

That stall is not primarily about skill. It is about permission, clarity, and fear.

The workers closest to the processes that need to change are often the ones with the most hesitation about using AI on those processes. They are protecting institutional knowledge they have built over careers. They are uncertain about what AI gets right and what it gets wrong. They have watched enough technology initiatives fail that skepticism feels like wisdom. And in most organizations, no one has given them a clear mandate to experiment, a safe environment to fail, or a framework to convert their expertise into AI capability.

The real competitive moat in the AI era is not the organization with the best AI platform. It is the organization that has most effectively converted its domain expertise into persistent, scalable AI capability.

This is the leadership failure hiding inside the utilization gap. It is not that workers lack access to AI. The Microsoft Work Trend Index found that 75 percent of knowledge workers already use AI at work, and 78 percent bring their own tools when their organizations don't provide them. Workers are finding the tools. Organizations are not giving them the authority, the architecture, or the accountability to build with them.

That conversion does not happen top-down. It happens when the people closest to the work are empowered, equipped, and trusted to lead it.

WHAT ACCELERATION ACTUALLY LOOKS LIKE

The distinction between adoption and acceleration is visible in the output, not the input.

An organization in adoption mode asks: How many people are using AI? An organization in acceleration mode asks: What new capabilities have we built in the last 90 days that we couldn't have built before?

Adoption produces reports on license utilization. Acceleration produces new decision frameworks, new workflow architectures, new tools built by domain experts who previously had no way to build them. Adoption is measured in usage. Acceleration is measured in capability creation.

The OpenAI enterprise research confirms the gap is structural. Frontier firms — the organizations actually pulling away from peers — send twice as many messages per seat as the median enterprise and seven times more messages to custom AI workflows. They are not using AI more. They are using it differently. They have moved from query to build. From asking to creating. From productivity tool to capability engine.

That shift does not require a new AI platform. It requires a different leadership posture — one that treats every domain expert in the organization as a potential builder, not just a user.

THE LEADERSHIP SHIFT

Three questions to take back to your leadership team.

First: where in your organization is domain expertise deepest and workflow design most manual? That intersection is where your dormant AI capability is concentrated. Those are the people who should be building, not just using.

Second: what are the fear and friction points keeping workers closest to the process from experimenting with AI on real work? Remove those before you add more training. Training without authority is theater.

Third: in the last quarter, has your organization built any new capabilities it could not have built before AI — not just completed existing work faster, but created something genuinely new? If the honest answer is no, the utilization gap is costing you more than time savings. It is costing you competitive position.

The research is no longer ambiguous. The organizations capturing value from AI are not the ones who deployed it earliest or most broadly. They are the ones who empowered the people closest to the work to build with it — and gave those people the architecture, the accountability, and the leadership support to convert their expertise into something that compounds.

The dormant AI capability in your organization is real. It is sitting in every department. It is waiting on a leadership decision.

Capability without readiness is just exposure.

Scott Wise brings 30 years of transformation consulting experience to the most important leadership challenge of our time. Author of AI4Leaders: Amplify Your Impact and certified in AI by MIT and Oxford, he helps executive teams and organizations move from AI-Curious to AI-Capable. Explore his work at ScottWise.ai.