AI Training Should Change How People Work, Not Just What They Know
A lot of AI training looks the same. Get everyone in a room for an hour. Walk through what the tools can do. Maybe run a live demo. Send people back to their desks.
Recent research suggests that approach, on its own, produces almost no lasting behavior change.
McKinsey published research late last year examining why AI training efforts keep falling short. Only about a quarter of workers report receiving any training on how to actually collaborate with AI, and even when they do, the format rarely sticks. In one study of Microsoft Copilot adoption, seven out of ten participants ignored onboarding videos entirely.
PMI's analysis landed in the same place. Organizations are treating upskilling as a one-time learning event when it should function as an ongoing operational capability.
General Awareness Doesn't Produce Adoption
Awareness fades fast when nothing in your daily routine reinforces it.
AI training has the same problem. People leave understanding basic prompts. But then they open their laptops Monday morning, and every workflow is exactly the same as before.
McKinsey's research puts a finer point on this. Lasting adoption requires four things working together: people knowing what to do differently, believing in why it matters, feeling supported by leadership, and seeing reinforcement in the systems around them.
The missing ingredient in most AI training is workflow redesign. Without learning how to change the actual sequence of steps someone follows to complete a recurring task, training becomes trivia.
What Effective AI Training Looks Like by Department
Instead of training everyone on AI in general, customize training per department and redesign one workflow in each with AI embedded in it.
Sales team. The weekly pipeline review takes hours because reps manually pull data from the CRM, compile notes from calls, and build a summary for their manager. A redesigned workflow pulls deal data automatically, uses AI to summarize recent call notes and flag stalled opportunities, and generates a draft pipeline update. The rep reviews, adds context the AI missed, and submits.
Customer support. Agents spend most of their time reading tickets, searching the knowledge base, and typing out responses that are 80% similar to ones they've written before. A redesigned workflow has AI draft responses based on ticket content and prior resolutions. The agent reviews for accuracy, adjusts tone, and sends.
Finance. Monthly reporting involves pulling data from multiple systems, reconciling numbers, and writing variance explanations. A redesigned workflow automates the data pull and has AI generate draft variance commentary based on the numbers. The analyst reviews, corrects, and adds context.
People learn AI by using it inside real work, not by studying it in a conference room. When you stop thinking of training as a separate event and start thinking of it as workflow redesign, adoption takes care of itself.
How to Know If It's Working
Measurement matters here, but keep it simple. Track three things before and after the workflow change.
Time per cycle. How long does the task take now versus before? If there's no meaningful reduction, the redesign needs adjustment.
Output consistency. Is the quality of the deliverable holding steady or improving? AI should raise the floor, not lower the ceiling.
Adoption persistence. Is the person still using the new workflow after 30 days? If they've reverted to the old process, something in the design is creating friction.
You don't need a dashboard for this. A simple log works. The goal is to confirm that the workflow change produced a real, sustained shift in how someone works.
Training Follows Behavior, Not the Other Way Around
Most organizations start with training and hope behavior follows. The research says otherwise.
When you start with a single workflow change, something interesting happens. People learn AI by using it in context. That's the kind of learning that sticks. And it scales naturally, because once someone redesigns one workflow successfully, they start looking at the next one on their own.
. . .
Want to save hours each week by turning work into repeatable AI workflows?
The Fortune 100 AI Skills Library™ (https://www.technical-leaders.com/library) includes plug-and-play prompts built to save leaders time and money. Copy, paste, and edit in 60 seconds, then apply them across planning, execution, and reporting.
Top comments (0)