Everyone Has Been Trained in AI. So What Has Actually Changed?

2 days ago

9 min read

Imagine an organisation that has just completed a major AI training initiative. Hundreds of employees have attended the courses, the feedback has been encouraging, and management is pleased with the participation numbers. The learning team has delivered what was planned, and everyone has good reason to feel that something worthwhile has been achieved.

But three months later, someone asks a question that is a little harder to answer. What are our employees actually doing differently now, and has any of that made a meaningful difference to the way we work?

Perhaps some employees are using AI to draft emails, summarise documents or prepare presentations. A few enthusiastic individuals may have discovered clever ways to make their work easier, while others have barely touched the tools since attending the course. Managers, meanwhile, may still be struggling to identify which processes have improved or whether the organisation is making better decisions as a result.

None of this necessarily means the training was poor. Participants may genuinely have enjoyed the experience, learned useful techniques and left the classroom feeling confident about using AI. The difficulty is that learning something and applying it consistently in the workplace are two different achievements, and organisations sometimes assume that one will naturally lead to the other.

Perhaps the question isn't whether the AI training was successful, but whether it has helped the organisation develop a capability it didn't have before.

AI training is becoming a priority. But what happens afterwards?

A recent study offers an interesting perspective on this question. The State of Enterprise Learning 2026 report was commissioned by edX Enterprise, a provider of corporate learning and professional development, and conducted by NewtonX, a business research firm.

The researchers surveyed 507 learning and workforce-strategy leaders from organisations employing at least 500 people. While 83% said AI capability was a priority for their organisations, only 29% considered their organisations to have mature AI capabilities.

That's quite a difference between what organisations want to achieve and where they believe they currently stand. Of course, these figures reflect the assessments of the leaders surveyed rather than independently measured organisational performance. We shouldn't interpret them as evidence that the remaining organisations have failed.

Nevertheless, the findings raise a question worth considering. If AI has become such an important priority, are organisations paying enough attention to what happens after their employees have completed the training?

Perhaps we've become very good at organising learning opportunities, but haven't given quite as much attention to helping people put that learning into practice.

*AI priorities are high, but reported capability is lagging.*

Learning to use AI is only the beginning

In my training classes, I often see the excitement when participants discover what AI can do. Someone learns how to structure a prompt, provide better context and get a useful response from ChatGPT. Suddenly, a task that once seemed difficult becomes more manageable, and the possibilities begin to open up.

That's exactly what a good learning experience should achieve. But the more interesting question is what happens when participants return to work and try to apply those techniques to their own responsibilities.

Take a team responsible for preparing monthly management reports. After attending AI training, one employee discovers that AI can help produce an initial draft much faster. That's a useful skill, and it may save the employee considerable time. However, if the process remains dependent on that individual, nobody else understands how the output was prepared, and there is no agreed method for checking its accuracy, the organisation hasn't really changed very much.

Now imagine the same team taking the idea further. They examine how reports are currently prepared, identify where AI can genuinely help, and agree on how the results should be checked before reaching management. Other team members learn to follow the same process, and the manager reviews whether the reports are becoming more timely, accurate and useful.

In the first situation, an individual has learned to use AI. In the second, the team has begun developing a way of working that can be repeated, reviewed and improved.

That's what I mean by organisational capability. It isn't simply having employees who know how to use AI tools. It's having the people, working practices and support needed to produce useful results consistently, rather than depending entirely on one enthusiastic employee.

And if that's the outcome we're hoping for, perhaps we need to reconsider where the training conversation begins.

Perhaps we're starting with the wrong question

When organisations discuss AI training, the conversation often begins with the courses they want their employees to attend. It might be Prompt Engineering, Generative AI for Business, AI Productivity or something more specialised. These are all legitimate areas of learning, but I sometimes wonder whether we're beginning the conversation at the wrong end.

Before deciding what employees should learn, shouldn't we first understand what the organisation wants them to do better?

A reporting team might be spending too much time consolidating information from different sources. Managers may need help examining business problems and evaluating recommendations. Employees dealing with customer enquiries might be struggling to review lengthy documents quickly without overlooking important details.

Once we understand the work and the difficulties involved, it becomes much easier to decide where AI might be useful and what employees actually need to learn.

This also helps explain why sending everyone to the same course may not always produce the best results. A manager doesn't necessarily need the same AI skills as an analyst or someone working in corporate communications. The manager may need to evaluate recommendations and challenge assumptions, while the analyst needs to verify calculations and interpret results. A communications professional, on the other hand, may be more concerned with accuracy, context and tone.

All three may use the same AI tool, but the judgement required to use it well can be quite different.

Of course, there's nothing wrong with starting with a general AI course, particularly when employees are still unfamiliar with the technology. Building awareness and confidence is a worthwhile objective in itself. But when the intention is to improve a particular business process or decision, the learning needs to go further. Employees must understand how to apply what they've learned to their own work, and the organisation needs to support that change.

This is one reason we've always placed considerable emphasis on hands-on learning at FYT. Participants need opportunities to practise, make mistakes, question the results and understand why a particular approach works or doesn't work. Watching an impressive demonstration can create interest, but trying to apply the technique yourself is where much of the real learning happens.

Even then, the classroom can only take us so far. Real workplace problems are often more complicated than training exercises. The information may be incomplete, the question poorly defined, or the AI response surprisingly convincing despite being wrong. Employees need opportunities to work through these situations using appropriate tools and information, with guidance and feedback along the way.

And that's where the organisation itself has an important part to play.

The workplace has to be ready too

Consider an employee returning from an AI workshop with several ideas for improving an existing process. The employee is enthusiastic and wants to try something different, but the manager is uncertain about the benefits, the organisation hasn't clarified which tools are approved, and there are concerns about sharing internal information. With everyday deadlines already competing for attention, experimenting with a new approach quickly becomes a lower priority.

Eventually, the employee returns to the familiar way of working, not necessarily because the training was ineffective, but because the environment wasn't ready to support the change.

This is where managers and organisational leadership become particularly important. Employees need access to suitable tools, clear guidance on responsible use and opportunities to experiment without creating unnecessary risks. Just as importantly, managers need to understand why a different approach might be worthwhile and be prepared to review the results.

The edX/NewtonX study found that 78% of the organisations surveyed lacked a strong combination of leadership support, clear direction, resources and the ability to deliver their workforce development plans. The researchers also identified a smaller group of organisations reporting more mature AI capabilities that were considerably more likely to have these conditions in place.

Although this relationship doesn't establish cause and effect, it reinforces something worth remembering. We can develop individual skills through training, but changing how an organisation works requires more than individual enthusiasm.

There is also the question of employees who are less comfortable with AI. Not everyone is excited about new technology, and some may be concerned about making mistakes, uncertain about the benefits or simply too busy to experiment. Rather than asking them to use AI because it's becoming important, it may be more productive to begin with a problem they already face and explore whether AI can help.

When people can see a practical reason for learning something, and know they'll receive support in applying it, the conversation becomes much easier.

How would we know whether anything has improved?

This brings us to something that I think deserves more attention in conversations about AI training: how we measure success.

Attendance and participant feedback are useful, and as a trainer, I certainly want people to enjoy the class, understand the concepts and leave feeling confident enough to try what they've learned. But these measures tell us primarily about the learning experience. They don't necessarily tell us what has happened in the workplace afterwards.

Suppose employees have learned to use AI when preparing business reports. Rather than simply asking whether they're using the tool, we could examine whether reports are being completed more efficiently, whether fewer corrections are required and whether managers find the analysis more useful.

We might also discover benefits that aren't easily captured by time savings. Perhaps employees are asking better questions, identifying inconsistencies earlier or becoming more careful about checking information before making recommendations. These changes may be just as valuable as producing a report ten minutes faster.

We should also be cautious about using the frequency of AI usage as evidence of capability. Someone who submits twenty prompts a day isn't necessarily doing better work than someone who uses AI only occasionally but knows exactly when it is useful, when the output needs verification and when human judgement matters more.

And even when AI saves time, the value depends partly on what happens to that time. If an employee saves fifteen minutes preparing a report but still has to spend another twenty minutes correcting errors, the apparent efficiency gain may not be much of a gain at all.

None of this requires an elaborate measurement system. Organisations can begin with a few relevant questions about a particular workflow, establish how the work is currently performed and review what has changed after employees have had time to apply their learning.

The results won't always be straightforward, and we shouldn't attribute every improvement directly to training. But a thoughtful review gives management a much stronger basis for deciding whether to continue, adjust or expand the initiative.

*Workplace improvement is the real test of capability.*

Building capability doesn't have to begin with a big initiative

For organisations that are only beginning their AI journey, all this may sound like a great deal of effort. But it doesn't need to start with a large transformation project or an expensive organisation-wide program.

A more manageable approach is to choose one area where employees face a genuine difficulty, understand how the work is currently done and identify where AI might help. Provide the relevant learning, establish sensible boundaries and give people the opportunity to experiment with real tasks.

After a reasonable period, review the results with the employees and managers involved. If the approach is working, refine it and consider whether other teams could benefit. If it isn't, investigate why before investing further.

This also provides a practical response to the concern that AI tools are changing too quickly for organisations to keep up. Specific features and interfaces will certainly evolve, but the ability to define problems, interpret information, verify outputs and exercise sound judgement will remain useful regardless of which tools eventually become dominant.

In other words, organisations shouldn't focus only on teaching employees how to operate today's AI tools. They should also help people develop the thinking and working practices that will allow them to adapt as the technology changes.

What should we celebrate after AI training?

I certainly don't think organisations should stop investing in AI training. Helping employees understand these technologies, recognise their limitations and use them responsibly is an important part of preparing for the future.

But perhaps we need to be a little more careful about what we celebrate.

High participation, positive feedback and successful course completion are achievements worth recognising. They tell us that people have been given opportunities to learn and, hopefully, have developed skills they didn't have before.

What they don't automatically tell us is whether the organisation has become better at doing anything. That requires looking beyond the classroom to see whether employees are applying what they've learned, whether managers are supporting those efforts and whether the resulting changes are genuinely useful.

So the next time an organisation completes an AI training initiative, I'd encourage management to revisit it a few months later and ask a different question.

What can our organisation now do reliably, responsibly and better than before?

Because when we can answer that with real workplace examples, we're no longer talking only about AI training. We're beginning to talk about AI capability.