Critical Thinking in the Age of AI
Keeping your brain in charge
Artificial intelligence is changing not only what we can accomplish at work, but also the way we think while accomplishing it.
Across organisations, professionals are increasingly using AI to draft reports, analyse information, structure ideas, solve problems, prepare presentations and summarise large amounts of material. The benefits are obvious: tasks that once took hours can sometimes be completed in minutes, and the resulting work can appear remarkably polished.
But there is another side to this transformation that deserves more attention.
When a system can produce a convincing answer almost instantly, we may become less conscious of the thinking that happens between a question and a decision.
The risk is not necessarily that AI will think instead of us. The more subtle risk is that we may gradually stop noticing when we are thinking less.
We accept a summary because it is convenient.
We trust a recommendation because it sounds plausible.
We adopt a formulation because it is more articulate than the one we would have produced ourselves.
And sometimes, we mistake fluency for accuracy.
This raises an important question for anyone working with AI:
How do we continue to exercise judgment when intelligent systems increasingly participate in our thinking?
Critical thinking is something we practise
Discussions about critical thinking and AI often move towards extremes.
One response is to become suspicious of technology and try to verify everything. Another is to introduce increasingly elaborate rules, governance frameworks and training programmes.
Neither approach captures what I find most useful in everyday professional practice.
I prefer to think about cognitive fitness.
Just as physical strength is maintained through regular movement rather than occasional heroic effort, our capacity for judgment can be strengthened through small, repeated acts of conscious thinking.
A machine can assist a movement without taking the movement away from us completely. The same principle can apply to AI.
The question is not whether we use the tool.
The question is how much of the thinking we remain actively involved in.
AI can help us move faster, explore further and see possibilities we might otherwise miss. But the human contribution becomes particularly important when we need to interpret, question, contextualise, decide and take responsibility.
That is where agency comes in.
The five-second pause
Consider a familiar workplace situation.
You ask an AI system to prepare a strategic summary. A few seconds later, you receive a coherent, well-structured response.
It sounds right.
And because it sounds right, your brain may be inclined to experience it as right.
This phenomenon is related to processing fluency: information that is easy to process can feel more familiar, credible or true, even when fluency itself provides no guarantee of accuracy.
This is where a very small intervention can make a meaningful difference.
Pause.
Not for ten minutes. Not necessarily for another elaborate verification process.
Sometimes, five seconds is enough to introduce a little cognitive friction.
Ask yourself:
Does this sound right – or is it actually right?
That question changes your position in the interaction. Instead of simply receiving an answer, you begin evaluating it.
You may notice an unsupported assumption.
You may spot information that is missing.
You may realise that the answer reflects your original framing too closely.
You may recognise that the system sounds more certain than the evidence warrants.
This is the kind of deliberate mental effort associated with what Daniel Kahneman described as System 2 thinking: slower, more effortful and analytical processing, as opposed to the fast, automatic responses of System 1.
The objective is not to operate in System 2 all day. That would be exhausting and unnecessary.
The objective is to know when to engage it.
Three habits for cognitive fitness
In my work as a change manager, coach and learning and development professional, I have found it useful to translate this idea into three simple practices:
1. NOTICE: become aware
Start by noticing what happens in your own thinking when you work with AI.
What makes you trust an answer?
Is it the speed?
The confident tone?
The polished language?
The fact that it confirms what you already thought?
Awareness comes before change.
One useful test is to see whether you can explain an idea in your own words without leaning on the AI-generated formulation. The principle behind the well-known Feynman Technique (attributed also to Einstein) is particularly relevant here: if you truly understand something, you should be able to explain it simply.
Clarity is therefore not just a communication skill. It can also be a test of understanding.
2. CHOOSE: exercise agency
Not every task needs the same degree of human involvement.
Sometimes delegating a task almost entirely to AI is perfectly reasonable. At other times, the context, consequences or ambiguity mean that human judgment needs to remain firmly in the driver’s seat.
The important thing is to make that decision consciously.
Ask yourself:
What am I comfortable delegating? What do I want to remain responsible for?
Your experience, context, values, professional judgment and ethical responsibility cannot simply be transferred because a system can produce an answer.
AI can be a collaborator.
It does not have to become the authority.
3. SHARPEN: use AI to challenge your thinking
There is another possibility that is often overlooked.
Instead of asking AI only to give us answers, we can ask it to make our thinking better.
Ask it to challenge your assumptions.
Offer a counterargument.
Generate a perspective you have not considered.
Identify weaknesses in your reasoning.
Point out what information might be missing.
Simulate how someone with a different role or interest might interpret the situation.
Used this way, AI becomes less of an answer machine and more of a thinking partner.
The paradox is interesting: the better we learn to use AI, the more deliberately we may need to exercise the capabilities that make us human thinkers.
Small repetitions matter
Critical thinking does not need to become another productivity burden.
In fact, I believe the opposite.
The most sustainable approach may be to build small moments of reflection into the way we already work.
A pause before accepting an answer.
A question before delegating a task.
A challenge to our own assumptions.
An attempt to explain something without AI.
A deliberate decision about where human judgment needs to remain.
These moments may seem insignificant individually. But repeated over time, they can become habits.
The questions I return to are simple:
- What am I thinking here?
- What am I asking AI to do for me?
- And what do I still want to keep as mine?
From an observation to the AI Thinking Gym
Since 2024, I have been exploring the role of AI more systematically in my work across coaching, learning design, methodology development and organisational development.
The more I worked with these systems, the more I became interested in a question beyond productivity:
What happens to our thinking when AI becomes part of the way we work every day?
This question eventually led me to develop the AI Thinking Gym.
The idea grew from several strands of my professional practice: coaching approaches and competencies, positive psychology, solution-focused practice, learning and development, education, and my broader interest in how sustainable behavioural change happens.
It also reflects something I have increasingly come to value: consistency tends to be more powerful than intensity.
We do not become physically fit through one demanding workout. We build fitness through repeated practice.
Why should cognitive fitness be fundamentally different?
The AI Thinking Gym translates this idea into a lightweight 21-day practice for thinking well with AI.
It is designed to help professionals:
- notice cognitive shortcuts and automatic reactions;
- make more intentional decisions about delegation;
- practise questioning and reflection;
- strengthen active judgment;
- use AI to expand rather than replace their thinking.
The experience consists of short, one-to-two-minute daily exercises organised around three progressive phases:
NOTICE. CHOOSE. SHARPEN.
Each day offers a brief scenario, reflection or brain insight designed to create a small moment of cognitive exercise.
There is no score to achieve and no performance evaluation. The tool is designed as a reflective coaching experience rather than a test. It does not store reflections remotely; they remain on the device you use.
It is also deliberately a work in progress. I am continuing to refine the experience based on what I learn from people who use it.
You can try the AI Thinking Gym here:
https://ai-thinking-gym.lovable.app
Keeping the human in the loop
AI will continue to become more capable.
That makes the question of human agency more important, not less.
The future of work is unlikely to be about choosing between human intelligence and artificial intelligence. It will increasingly be about learning how to work with both — while understanding where each is most useful.
We should not be afraid of using AI.
But neither should we become so comfortable with its convenience that we stop exercising the capacities we ultimately remain responsible for.
Curiosity.
Judgment.
Creativity.
Context.
Discernment.
Wisdom.
Perhaps the goal is not to keep AI out of our thinking.
Perhaps it is to stay consciously in the loop.
The tools may become better at helping us navigate the map.
But we should remain responsible for deciding where we are going, and why.