The question at the heart of AI transformation that organisations cannot afford to overlook.
A forest can grow quickly without becoming resilient.
Rapid growth may be visible above ground, but the strength of a forest depends on less obvious things: the health of its soil, the diversity of its species, the relationships between organisms and its ability to regenerate when conditions change.
I have been thinking about this distinction recently in relation to artificial intelligence and the way organisations approach workforce transformation.
We are becoming very good at measuring what AI helps us produce. Faster analysis, shorter processes, improved efficiency and greater output are increasingly familiar markers of success.
But I wonder whether we are paying enough attention to what is happening beneath the surface …
Are we becoming more capable through our use of AI, or simply more productive with its assistance?
The difference may prove fundamental to the long-term health of organisations.
The human reality behind AI adoption
A recent Microsoft UK study published in October 2026 revealed a striking difference between how business leaders and employees view the country’s AI future.
While 84% of senior business decision-makers were confident that Britain could become a world-leading AI economy within five years, only 41% of employees shared that confidence.
The research, based on responses from 5,000 UK employees and 1,000 senior business decision-makers, also found that among employees already using AI at work, 71% reported a positive impact on their work or career, while 35% expressed concerns about job security.
These findings capture a tension I believe many organisations are still learning to navigate.
People can recognise the benefits of AI while remaining uncertain about what it means for their professional future.
And while Microsoft’s survey primarily measures confidence in Britain’s AI prospects, rather than individual workplace readiness, it raises a broader question about how technological change is experienced at different levels of an organisation.
From a leadership perspective, it is tempting to interpret hesitation as a lack of skills, awareness or openness to innovation.
Sometimes it is. But perhaps not always.
People may be asking more fundamental questions. Will my experience remain valuable? How will my role evolve? Am I being encouraged to develop, or simply expected to adapt to decisions already made?
These concerns cannot be resolved through AI training alone.
They require honest communication, meaningful involvement and a credible commitment to development.
In my experience, whether working with professionals, students or business teams, people engage more confidently with change when they understand its purpose and feel they have some agency in shaping their response.
This is why I increasingly see trust not as a desirable by-product of transformation, but as one of its essential conditions.
When better output does not necessarily mean better judgement
One of the most thought-provoking ideas in my recent reading came from Gleb Tsipursky’s October 2026 article in Training Journal, Is It Time to Rethink How L&D Measures AI Success?
Tsipursky challenges conventional approaches to evaluating AI-enabled learning and proposes a greater focus on how quickly employees develop genuine proficiency, including their ability to verify information, handle exceptions and exercise independent judgement.
The argument raises an important possibility: employees may become more efficient without necessarily developing the expertise required to make sound professional decisions.
Consider a junior marketing analyst preparing a market assessment.
Not long ago, this would involve gathering information, comparing sources, testing assumptions, recognising inconsistencies and gradually learning what makes an insight commercially meaningful.
Today, AI can produce a polished analysis in minutes.
There is considerable value in this. Technology can remove unnecessary effort, make sophisticated analytical approaches more accessible and create time for higher-value work.
But what happens to the learning that previously occurred through the process itself?
If someone repeatedly receives convincing answers without understanding how to interrogate them, they may become highly effective at producing work while remaining less confident in evaluating its quality.
The challenge becomes particularly visible when AI is wrong.
Would the analyst recognise a questionable assumption? Could they explain why a recommendation might be misleading? Would they know when to challenge the output rather than accept it?
Of course, professional competence has never depended on doing everything manually. We have always developed expertise alongside tools.
The question is not whether AI should take over routine activities. It is what happens when we are deliberately replacing the learning opportunities that disappear when those activities are automated.
It is what happens when we are deliberately replacing the learning opportunities that disappear when those activities are automated.
This is a challenge I recognise from higher education.
As AI becomes part of how students research, analyse and communicate, the quality of a final submission tells us less about the understanding behind it.
We therefore need to pay greater attention to the thinking, judgement and decisions that learners can demonstrate.
The same principle now applies in the workplace.
My point: A sophisticated output is not necessarily evidence of sophisticated understanding.
The learning question organisations need to ask
This tension becomes even more interesting when considered alongside recent research into the maturity of corporate learning and development.
The 2026 State of L&D Maturity Report by 360Learning, based on 1,062 organisational self-assessments collected between February 2025 and July 2026, identifies learning needs analysis and impact measurement as significant areas of weakness among participating organisations.
According to the report, 52% of respondents identified learning needs analysis as their weakest area, while 46% highlighted measuring learning impact.
As the findings are based on voluntary self-assessment, they should not be interpreted as representative of the entire L&D profession. Nevertheless, they point towards an important strategic challenge.
Many organisations have developed sophisticated ways of delivering learning.
They are less certain about identifying what people genuinely need to learn and evaluating whether that learning changes how they perform.
We can count course completions, attendance, engagement and satisfaction. These measures may tell us whether a learning experience reached its audience.
They tell us much less about whether people can apply what they learned when faced with a difficult decision.
AI makes that distinction even more important.
An employee may adopt AI enthusiastically and complete tasks significantly faster. Yet speed alone cannot tell us whether they have become better at evaluating information, navigating ambiguity, identifying risk or making sound decisions.
Drawing on the dimensions suggested by Tsipursky in Training Journal, a more meaningful assessment might explore how well people recognise errors, manage exceptions, explain their reasoning and transfer knowledge to unfamiliar situations.
These are promising measures to explore, rather than a universally established framework for assessing AI-enabled professional competence.
This also changes the role of learning professionals.
Rather than beginning with the question, “What training should we provide?”, we need to become more comfortable asking, “What is preventing people from performing effectively, and what would meaningful improvement look like?”
Sometimes the answer will involve new knowledge or skills.
Sometimes it will involve organisational processes, management practices, role clarity or the opportunities people are given to exercise judgement.
Learning strategy becomes more valuable when it helps organisations understand these distinctions.
What happens when we remove the conditions for growth?
This brings me back to the forest.
A resilient ecosystem does not develop through growth alone. It develops through continuous interaction with its environment, adaptation and regeneration.
Professional expertise has a similar quality.
We rarely develop good judgement simply by receiving information. We develop it by testing our understanding against reality, experiencing uncertainty, receiving feedback and learning to recognise patterns over time.
Think about how people traditionally progress through a profession.
Early-career employees often begin with work that more experienced colleagues consider routine. They draft reports, examine data, prepare recommendations and deal with relatively straightforward problems.
Much of this work is precisely what AI can now help automate.
Yet these tasks also provide the foundation on which more advanced judgement is built.
They allow people to notice what matters, understand professional standards, learn from mistakes and gradually develop the confidence to take responsibility for increasingly complex decisions.
This is where I believe the concerns about AI-enabled proficiency raised by Tsipursky have implications beyond learning measurement. They also invite us to reconsider how expertise itself develops within organisations.
If we automate these experiences without considering their developmental purpose, we may inadvertently weaken the pathways through which future expertise emerges.
I don’t argue for protecting inefficient work for the sake of tradition. I simply say we may need to rethink how people learn through work.
AI-generated analysis could become the starting point for critical discussion rather than the final answer. Simulations could expose employees to difficult scenarios before they encounter them in practice. Managers could create opportunities for teams to explain, challenge and reflect on AI-supported decisions.
Used thoughtfully, AI could accelerate the development of expertise rather than diminish it.
But that outcome will not happen automatically. It requires intentional learning design and leadership attention.
From AI adoption to organisational capability
I believe this is where the conversation about AI transformation needs to evolve.
Introducing technology and developing organisational capability are related, but they are not the same undertaking.
Technology can often be deployed at speed. Human capability develops through a more complex combination of experience, learning, confidence, relationships and organisational culture.
A successful AI strategy therefore needs more than technical infrastructure and adoption targets.
It needs an equally thoughtful approach to the people who will work with these systems, question their recommendations and remain accountable for the decisions that follow.
This broader question of capability is also emerging at a policy level.
In its October 2026 publication, AI Sovereignty: Preserving UK Agency in an AI-Enabled World, the UK’s Council for Science and Technology argues for strengthening the country’s ability to develop, understand and govern AI, rather than relying solely on purchasing and deploying technologies developed elsewhere.
Although the report addresses national technological capability, I see a parallel at organisational level.
There is a difference between having access to powerful systems and developing the knowledge, confidence and judgement to use them responsibly.
This is particularly important in sectors where trust, risk and professional responsibility are central to performance.
Employees need to understand not only what AI can do, but also its limitations and the circumstances in which human intervention is essential.
Leaders need to consider whether their organisations are creating opportunities for people to deepen their expertise, or simply encouraging them to become faster users of technology.
And learning professionals need to demonstrate impact beyond participation and completion, connecting development more clearly to judgement, behaviour and performance.
These are not separate concerns.
They are interconnected elements of an organisation’s ability to adapt.
A forest’s resilience depends on the strength of its underlying ecosystem, not simply the visible growth of individual trees.
Similarly, organisational resilience depends on more than the performance of individual tools or processes. It rests on the capabilities, relationships and judgement that allow people to respond when conditions change.
Perhaps we need a different measure of success
When leaders discuss AI, one of the first questions is often how much time it can save. It is an entirely reasonable question. But perhaps it should not be the only one.
I would like to see us become equally interested in asking:
What are our people becoming more capable of doing because AI is now part of their work?
- Are they making better decisions?
- Are they developing stronger analytical and critical thinking skills?
- Are they more confident in handling unfamiliar situations?
- Are they learning to recognise when technology should be challenged?
- And are we creating the conditions in which the next generation of professionals can develop the depth of expertise that organisations will continue to need?
These questions are harder to answer than calculating time saved or output generated.
But they may tell us much more about whether transformation is creating lasting value.
Throughout my career, working across education, professional development and business, I have seen how meaningful growth depends on more than access to knowledge or sophisticated tools.
It depends on the conditions we create for people to learn, experiment, contribute and gradually develop greater responsibility.
AI gives us remarkable new possibilities to support that process.
But it also challenges us to become more deliberate about the human capabilities we want to preserve and strengthen.
Perhaps the most important question for organisations is no longer simply how effectively people are using AI. It is whether their use of AI is helping them become better professionals. Because sustainable progress is not just about accelerating what we can produce today. It is about regenerating the capabilities we will depend on tomorrow.
If you have read this far – thank you! Clearly, something in here captured your attention. So I guess my question to you is this:
As AI takes on more of our everyday tasks, how do we ensure that we are not losing the very experiences through which professional judgement develops?
Leave a comment below and let’s continue the conversation there. I would love to see your perspective.
Gabriella
