90% of Employees Say AI Training Was Useful—A Month Later, No One Uses It?

Deep thoughts on AI and aspirations —— ByteDance Deep Thinking Circle

A study on M365 Copilot usage behavior reveals striking numbers: 90% of participants acknowledge formal training was helpful, yet 70% never watch training videos. What actually gets them using it is hands-on trial and error, plus discussions with colleagues.

There’s an even more common scenario. A company rolls out AI literacy training for all employees. At course completion, everyone can articulate what generative AI can do. A month later, actual application is virtually nonexistent. The course itself had no quality issues—the problem lies outside the curriculum.

Both sets of facts point to the same conclusion, and it’s not gentle: training doesn’t change behavior. Companies hoping to drive AI adoption through a single training program often purchase just a month of psychological comfort.

Behavior Isn’t a Knowledge Problem, It’s a Structural Problem

Why do people attend training, understand it, then not use it? Because employee behavior has never been determined by knowledge—it’s determined by three harder things: what’s on the performance review, what managers do daily, and whether processes have room for AI.

Knowledge is the softest layer, poured in within a few hours. Behavior gets stuck at the hardest layer: performance metrics haven’t changed, so time saved using AI doesn’t show up in KPIs; managers don’t use it themselves, so AI-generated deliverables make employees look underutilized; processes haven’t adapted, so AI outputs must be forced back into approval and reporting workflows designed for manual work. When structure doesn’t move, acquired knowledge naturally falls away, and people return to the path that was previously rewarded.

Background data reveals how widespread this is. In McKinsey’s global AI survey, 78% of organizations are already using AI in at least one business function; another report on the future workforce shows 75% of US workers expect AI to change their roles within five years, but only 45% have actually upskilled recently. That gap has never been filled by scheduling classes—it’s filled by moving learning from the classroom into the cracks of organizational structure.

I have another uncomfortable observation: for many companies, the real buyer of AI training is management anxiety. The budget purchases peace of mind that “we did something.” Judging whether training serves this purpose is simple—look at what else happens in the organization after training ends. If nothing happens, what was purchased was reassurance.

Change Has Three Depths, Vastly Different Price Tags

Viewing AI skill-building as a change management project, you can distinguish three layers by depth of transformation. Each layer is more expensive than the last, and each gets closer to actual returns.

First layer: basic awareness. Get employees to understand AI language, tools, and risks; remove hesitation and encourage experimentation. Most companies invest most here because it’s visible and measurable—classes generate attendance records. But this layer only addresses willingness to touch AI, not actual usage. It explains why this layer most easily produces attractive event photos and the ugliest business results.

Second layer: embed AI into processes and performance metrics. Contrast the company whose training had no effect with another that did four things simultaneously: directly integrate AI into workflows; have managers lead by example and use it themselves; redesign performance metrics to incorporate continuous online learning time, scenario development, and efficiency KPIs; build a peer-led support community. Same courses, but both awareness and application rose. The difference wasn’t course quality—it was the supporting infrastructure. Notice this layer begins touching power and interests: changing metrics means redefining what constitutes good performance; managers leading by example means publicly exposing their own learning curves. Most companies stop at the first layer—not because they don’t know the second is valuable, but because leadership resolve is only truly tested starting at the second layer.

Third layer: rebuild the business itself with AI. Develop AI application scenarios specific to your business domain that create competitive advantage, connecting skill-building directly to business transformation. Few companies reach this layer, but performance gains happen precisely here—the first two layers are merely admission tickets.

Before Launch, Pass Four Self-Check Questions

If you can’t answer any of these four questions, hold off on the training budget. Training has never been the starting point of transformation—performance metrics and manager behavior are. Courses just plant seeds in soil that’s already loosened. Without loose soil, scattering more seeds only creates photo opportunities.

There’s a positive example to reference. When McKinsey rolled out its internal generative AI platform Lilli, it didn’t take the “distribute accounts and schedule classes” route: leadership repeatedly articulated the transformation story, launched new capabilities and processes in parallel, developed learning content, provided practical support, and operated skill-building as a change management project. The result: 3,000 users, time required to gain insights reduced by 20%. Notice the structure of this case: story, process, support—all present. Accounts and courses were actually the least important parts of the entire initiative.

When You Don’t Need the Full Production

The boundaries of this approach need clarification. For organizations with hundreds to thousands of people, where behavior is deeply locked by structure, it’s worth establishing as a change management project led by the CEO, not HR—because it touches performance metrics, processes, and manager behavior, all beyond HR’s authority. For teams of ten or so, it’s completely unnecessary. The CEO using AI daily and producing visible results is itself the most efficient change management project. Employees’ observation of leadership behavior runs deeper than trust in any curriculum.

One more thing to say honestly: this approach assumes leadership genuinely wants to embrace change. If management’s calculation is merely using AI to cut costs and staff, employees will quickly sense it, and no amount of reskilling design will elicit sincere engagement. Framing learning as a long journey the organization takes with you versus dumping fear on employees—the output difference between these two narratives far exceeds any difference in course quality.

Before approving the next AI training budget, do one thing that takes five minutes: pull out the company’s performance review form and read it. If nothing on it rewards new ways of working, what that budget will likely purchase is next month’s peace of mind.