Today, more and more companies are buying AI software licenses and running training courses to introduce them, hoping to keep pace with new technologies. Often, the main metric tracked is simply the course completion rate. What isn’t measured, however, is whether people are actually working any differently.
Data suggests it is worth asking the question: are we truly changing how we work? A February 2026 Gallup survey of over 23,000 U.S. employees reveals that while the growing presence of AI is beginning to reshape workplace dynamics, evidence of substantial changes in how organizations actually get work done remains limited.
Is Understanding How AI Works Enough?
Most AI training is designed like any other course: explain the key concepts, discuss what generative AI is, show how to write a good prompt, and wrap up with a quiz. It’s a reasonable starting point, but it treats AI strictly as a topic to learn rather than a skill to acquire and integrate into daily work.
According to findings from Microsoft’s 2026 Work Trend Index, a growing share of workers are using AI in ingenious ways: in many cases, people are ready to take a step forward, but the systems built around them are not.
The barrier is rarely tool access or learner motivation – it lies in the dynamics, context, and ways the tool is embedded into the actual workflow. The quality of the course certainly matters, but the context in which it lands matters even more.
Start with Tasks
Instead of teaching the same tool to the entire company in a one-size-fits-all manner, it is far more effective to identify recurring tasks for each role where AI can make a genuine impact. For example: HR writing job descriptions, customer service responding to repetitive tickets, or finance comparing data across different sources. Each of these represents a concrete, observable task, and each suggests a different way of using the same tool.
We find this exact logic in Action Mapping – the instructional design model created by Cathy Moore – which flips traditional training on its head: it starts with a measurable business goal and works backward through the observable actions people need to perform to achieve it, the practical activities needed to practice them, and only then the information required to do so.
Applying this concept to AI means we must stop asking “what do learners need to know about this tool?” and start asking “which tasks need to change, and how will we recognize when they are changing?”
Don’t Separate Training from Work
A one-hour training session completely isolated from daily work is easy to complete, but just as easy to forget. That’s because people need to apply what they learn while they learn it.
The 5 Moments of Need model, developed by Bob Mosher and Conrad Gottfredson, identifies five situations where a person needs to learn: when facing something for the first time, when wanting to learn more, when applying what was learned, when adapting to change, and when solving a problem. The first two benefit from structured training; the other three require workplace support delivered via just-in-time information and micro-learning.
Measure Skill Transfer, Not Just Completion
The most critical question isn’t the course completion rate, but how many people are consistently using the tool, on which tasks, and above all, with what performance outcomes compared to before.
There is a final element that decades of research consistently highlight. A meta-analysis of 89 studies confirmed that training transfer largely depends on a supportive work environment. AI data tells the same story: according to BCG’s AI at Work survey, the proportion of employees with a positive attitude toward generative AI jumps from 15% to 55% when strong leadership support is present.
Managers who use the tools themselves, discuss them in team meetings, and create space for workplace practice aren’t an optional extra, they are the prerequisite for training to work.
From Course to Ecosystem
The training course becomes just one component – a small mechanism within a broader ecosystem that encompasses task-based support, manager involvement, and the ability to track ongoing results.
It is a question of design rather than content: how training connects to the work people do, the platforms they already use, and the metrics that demonstrate what is working.
At APPrendere, this is what we do: build learning ecosystems where training is tied to real tasks, supported within the workflow, and measured by actual improvements.
Planning an AI learning journey for your organization? Let’s talk: https://www.apprendere.com/contact/
