For many leadership teams, buying AI has become easier than getting lasting value from it. McKinsey’s 2026 global survey found that only 37 percent of respondents attributed any EBIT impact to AI, while just 6 percent qualified as AI high performers. At the same time, Gallup found that 47 percent of U.S. employees said their organizations had integrated AI tools by mid-2026. The gap between deployment and measurable business results has become the central AI adoption problem.
I recently interviewed Rich Fox, President and CEO of Ardent Learning, and Bianca Baumann, its Vice President of Learning Solutions & Innovation. Their roles are part of Ardent’s current leadership team, and the company has built its AIRE framework around AI readiness and enablement. Their argument is straightforward: organizations get stuck when they treat tool access and training as the finish line, then leave employees and managers to figure out how AI should change actual work.
Measure Changed Work, Not Tool Activity
Fox sees one problem growing as AI gets easier to use: companies can generate more output without becoming more productive. “What’s missing is the discipline that just because you can do something with AI doesn’t mean you should,” he told me. More analyses, presentations, and dashboards can create the appearance of activity while making priorities harder to see.
Baumann makes the same point from the measurement side. “Most companies measure access and activity: logins, seats provisioned, maybe a satisfaction score. None of that tells you whether the business changed.”
Her preferred test is concrete. Pick a high-value workflow and ask whether turnaround time, error rates, decision quality, or the amount of higher-value work changed after AI entered the process. Measure the workflow before rollout so there is a baseline. A separate McKinsey readiness study found that leaders were 5.3 times more likely to report enterprise value when AI-enabled workflows were redesigned than when workflows stayed unchanged.
That distinction lines up with Microsoft’s 2026 Work Trend Index, which found that organizational factors such as culture, manager support, and talent practices accounted for twice the reported AI impact of individual effort. The same McKinsey survey cited above found that nearly three-quarters of AI high performers had fundamentally redesigned workflows, compared with about one-quarter of other respondents. Tool usage matters much less than whether the work itself changes.
Managers Decide Whether Strategy Reaches Daily Work
Senior leaders often believe employees resist AI. Gartner found something more complicated: 65 percent of employees said they were excited to use AI at work, yet 37 percent said they did not use AI even when they could because coworkers were not using it.
That makes adoption partly social. Employees watch colleagues and managers to determine what is normal, safe, and worth the effort.
Fox puts managers at the center of that process. He recommends that CEOs and CHROs stop asking only about query counts, token usage, or training completion. Instead, ask managers what should be happening with AI that still is not happening, how managers use AI themselves, and what employee questions they cannot answer.
“Managers are the hinge,” Fox said. “If they’re unequipped, nothing else matters.”
EY found a similar communication divide in its agentic AI workforce research. About 21 percent of employees below the VP level said their company had not clearly communicated its AI-agent strategy, compared with 9 percent of VP-level leaders. Workers at organizations with clear communication were also much more likely to report productivity gains from AI agents. The message from the top has to survive contact with the middle of the organization.
Fear Changes How Employees Use the Tool
Leaders also misread resistance when they reduce it to a skills problem. Fox hears employees asking whether AI will eliminate jobs, reduce the value of their expertise, or create broader harms. Those concerns shape behavior even when employees can technically use the tool.
“You can’t train your way out of fear,” Fox said. Telling employees that AI will support them carries little weight if managers cannot explain how roles, expectations, and career paths will change.
Microsoft’s Work Trend Index found that 65 percent of AI users feared falling behind if they failed to adapt quickly, while 45 percent said it felt safer to focus on current goals than to redesign work with AI. Employees can feel pressure to adopt and pressure to avoid risk at the same time.
That tension also makes ongoing learning necessary. The World Economic Forum expects 39 percent of workers’ core skills to change by 2030. Randstad’s 2026 Workmonitor found demand for prompt engineering as a core skill across functions rose 403 percent during 2025. PwC’s 2026 AI Jobs Barometer similarly found that skills in the most AI-exposed jobs are changing more than twice as fast as skills in the least exposed jobs. Training cannot be a one-time launch event when both tools and job expectations keep moving.
Start Narrow, Redesign, and Prove the Result
The practical answer is to narrow the scope. Baumann recommends choosing three to five priority workflows and asking employees where they reach for AI and where they avoid it. Ask managers what has changed in those workflows compared with a year ago. Listen for specific behavior: an employee can explain where AI changed an output or decision, and a manager can describe how they reinforced good use or corrected weak use.
This approach also helps leaders move faster because it limits the number of variables. A company can define the intended outcome, assess readiness, clarify expectations, redesign a workflow, support managers and employees through the change, and then measure the result.
That sequence echoes Ardent’s six-stage AIRE model, but it also reflects a broader lesson from current AI research: sustainable value comes from changing the operating system around the tool.
For leaders working through workplace AI adoption, the most useful question may be surprisingly mundane: what work now happens differently because AI exists? If leaders cannot answer that with a few concrete workflows and measurable outcomes, adoption remains superficial.
Fox offered an even simpler cultural test. “A sign adoption is real is when people stop talking about ‘AI work’ as its own category and just talk about the work.” That is a better destination than a dashboard full of logins.
