1. The Fine Line Between Support and Dependence
AI has not been part of our everyday working lives for very long, yet few technologies have moved into work and daily life at such speed.
At first, it was mostly curiosity. People asked a few questions, generated a paragraph, translated a document or searched for information.
Then, very quickly, AI began moving deeper into the way we work: research, analysis, planning, writing, design, programming, system development and, increasingly, even decision-making.
But alongside this convenience, another phenomenon is quietly emerging and spreading just as quickly: dependence on AI as part of the thinking process itself.
What begins as support for research, analysis or idea development can gradually become something else. We start asking AI before thinking for ourselves. We wait for AI to suggest an option before forming our own view. In some cases, we even begin relying on AI to tell us what to think next and what to do next.
And this is where an important issue begins to emerge:
the line between being supported by AI and allowing AI to replace part of our own thinking is becoming increasingly fragile.
So where does the real issue lie — in AI itself, or in the way we choose to use it?
AI, after all, is neither inherently good nor inherently bad. The value it creates — or the consequences it produces — depends largely on the quality of our thinking, our ability to understand the problem, and the way we bring AI into our work.
2. When AI Expands Human Thinking
For people who already have a strong foundation of knowledge, experience and perspective built over time, AI can become a powerful catalyst for thinking and creativity.
The clearest feeling, for me, is that AI can open up an entirely new horizon for thought.
Ideas that were once constrained by the information available to us, the time required for research, or even the limits of our familiar ways of thinking can now be expanded almost instantly. One line of inquiry can lead to another, then another, revealing possibilities we may never have imagined at the beginning.
This becomes especially valuable for people who have spent years building professional knowledge and experience.
A single useful suggestion from AI can sometimes connect with layers of knowledge, judgement and experience that have been accumulated over many years. That connection can trigger a new idea — something that might otherwise have taken far longer to emerge through more traditional ways of working.
But this process is not as simple as asking a question and receiving an answer.
Sometimes there is only one small idea in an AI response that makes us pause.
It may be something we have never considered before. Or it may connect with an experience or perspective that has been sitting somewhere in our own thinking for years.
That idea may not even be correct.
But it can open up a direction worth exploring.
And that is where the real work begins: through a continuing sequence of questions used to test assumptions, challenge the approach, identify what may be missing and assess whether an idea can actually work in practice.
Sometimes one well-timed suggestion is enough to activate a whole chain of associations, experiences and ideas already within us. From there, we can see a new way of approaching the problem, connect it with what we already know, challenge it and continue developing the idea in directions we may never have anticipated at the outset.
Creativity, then, does not lie in simply taking an idea generated by AI and using it. Creativity lies in recognising what is valuable among many possibilities, connecting it with our existing knowledge, and developing it into something new.
Once the problem is clear, working with AI can also pull us surprisingly quickly into a state of deep concentration.
Each response may reveal something else that needs to be clarified, an assumption that needs to be tested, or a direction worth exploring further.
To keep moving, we have to hold on tightly to the thread of our thinking — following the issue closely, remembering the connections, comparing new information with what we already know, and continually deciding which direction is worth pursuing. A moment of inattention can be enough to lose an idea or break the logic that is beginning to form.
And when we catch hold of an idea that is genuinely valuable, the thinking often does not stop there.
We keep going deeper, connecting pieces of information, trying different approaches and gradually transforming a fragmented idea into a structure clear enough to be applied in practice.
This process can naturally lead to deep work: a sustained level of concentration in which our attention remains fully engaged with a problem complex enough to demand continuous thought.
And if the idea is compelling, the objective is clear and each exchange generates meaningful feedback that allows us to keep moving forward, flow can emerge almost naturally.
At that point, we become absorbed in the work. The more clearly we understand the problem, the more we want to continue. The more the structure of the idea begins to take shape, the less we want to step away from the thinking that is unfolding.
Deep work keeps us in the depth; flow keeps us in the movement. When the two come together, we can sustain a long chain of thought without feeling that we are forcing ourselves to concentrate.
Seen from this perspective, AI can expand the space in which we think, accelerate exploration and help us develop an idea from an initial intuition into an approach that is clearer, deeper and more applicable in practice.
But the real value of AI does not lie in how much thinking it can do for us.
It lies in how much ownership we are still able to retain over our own thinking.
3. When Support Starts to Become Substitution
The same qualities that make AI powerful can also become a source of risk when we stop maintaining control over the thinking process.
AI is very good at producing answers that are structured, coherent and convincing.
A well-presented response can easily create the impression that it is correct, complete and ready to use.
But a plausible answer is not necessarily a correct answer — and even a correct answer may not be right for the context in which it is being applied.
If we do not have enough knowledge to recognise what needs to be verified, which assumptions may have been overlooked, or which constraints have not been considered, it becomes very easy to follow the logic AI has constructed for us.
A single suggestion can quickly grow into a process, a framework, a plan, a set of criteria, a form, or even an entire system.
And the further we go, the more convincing everything can begin to appear.
If this pattern repeats often enough, our natural response to problems can gradually change.
A problem appears — ask AI first.
We need an idea — wait for AI to suggest one.
A decision has to be made — see what AI recommends.
Over time, the habits of questioning, analysing and checking for ourselves are used less and less.
The ability to form an independent view, challenge assumptions and exercise judgement can begin to weaken.
Without noticing it, we can move from using AI to support our thinking to depending on AI to form that thinking for us.
And if that becomes a habit, what is lost is not simply our ability to solve one particular problem.
What is at stake is our capacity for independent thought and our ownership of the decisions we make.
4. From Individual Judgement to Organisational Risk
The risk moves to another level when AI-generated outputs are no longer used simply as reference material, but are introduced directly into the way an organisation operates without sufficient validation.
A management model may be logically constructed, yet disconnected from the organisation's actual direction, objectives or context.
A process may be comprehensive on paper but poorly matched to how the work really operates.
Roles and responsibilities may be clearly defined but exceed the organisation's current capability.
A system may contain a wide range of functions yet still fail to address the problem the business actually needs to solve.
The concern, therefore, is not simply that AI may produce an answer that is incorrect.
The greater risk is that proposals which sound reasonable are introduced into the organisation without first being placed within an overall management logic:
Where is the organisation trying to go? Where is it today? What capabilities does it currently have? And what change is genuinely necessary?
No matter how advanced a solution appears, it will not create value if the organisation does not yet have the people, capability, data, resources or level of maturity required to operate it effectively.
The wider the gap between the desired model and the organisation's actual ability to execute, the greater the risk of failure, additional work and loss of control.
The real risk is not only that AI may propose a model that is wrong, or one that is too far removed from the organisation's context, current state and execution capability. The greater risk is allowing that model to become part of the way the organisation operates before its strategic fit and practical feasibility have been properly tested.
5. When Ownership Becomes a Core Capability
Looking at both sides, one thing becomes clear: AI does not create value or risk on its own.
What matters is how we think, how we frame the problem, and how well we remain in control of the way we work with it.
So is working with AI easy or difficult?
From my own experience, the answer is:
Very easy — when we know what we want.
And also:
Very difficult — when we do not yet understand what we are really looking for.
When the problem is clear, we can work with AI very effectively.
We know what is worth keeping, what needs to be challenged, which assumptions must be verified, which direction deserves deeper exploration, and when it is time to stop.
More importantly, we know what problem we are trying to solve and what outcome we actually want to achieve.
The opposite is also true.
If we do not yet understand the problem ourselves, working with AI can quickly become an endless information loop.
Ask the same question in a slightly different way and AI may offer a different answer.
Add one more condition and another direction appears.
Keep asking, and the number of options continues to grow — many of them sounding equally reasonable.
Eventually, the problem is no longer a lack of answers.
The problem is having too many answers and no longer knowing which direction is genuinely worth pursuing.
That is why the difficult part of working with AI is not generating more options.
It is preventing ourselves from being pulled too far away from the original objective, knowing which direction deserves further attention, and knowing when to stop.
Without that control, we can easily become overwhelmed by the very volume of information AI is able to generate: the deeper we go, the more choices we encounter — and the more choices we encounter, the easier it becomes to lose direction.
There is a great deal of debate about how far AI will replace human beings.
But for me, the more important question is not:
“How much work will AI eventually be able to do instead of us?”
It is:
“How must we adapt and develop so that we can work effectively alongside AI while still retaining ownership of our thinking and decisions?”
AI can search for information, analyse data, develop ideas, propose options and generate highly structured outputs in an extremely short time.
But people still have to understand the context.
We still have to identify the real problem.
We still have to see what is not captured in the data.
We still have to judge what is appropriate, accept or reject assumptions, and ultimately take responsibility for the decisions that are made.
For that reason, perhaps the most important capability in the age of AI is not how many tools we know how to use, how many prompts we can write, or how much work we can make AI perform.
What matters more is whether we have the quality of thought required to know what to ask AI to do, what to trust, what to reject, and what we must ultimately decide for ourselves.
In a world where answers are becoming faster, more abundant and easier to access, human value will increasingly lie in our ability to frame the right problem, understand the context, exercise judgement and remain accountable for the choices we make.
The more powerful AI becomes, the more important human ownership of thought becomes. Ultimately, AI should extend human capability — not replace our right to think, judge and decide.
← Return to Insights