In a world where AI plays an ever-greater role in our daily lives, critical thinking has become a vital managerial skill. The aim is not to question everything, but rather to adopt a discerning approach in order to make sound decisions and motivate one’s teams. How can we avoid the pitfalls of thinking – whether individual or collective – in order to achieve greater objectivity? How vigilant should we be regarding data generated by AI? Finally, how can we draw on collective wisdom whilst avoiding group bias? Solange Hémery, an expert in executive coaching, offers insights on cultivating a mindset of methodical scepticism to be integrated into managerial practices.
Every day, our brains make a huge number of decisions. Yet, according to the work of psychologist Daniel Kahneman, the majority of these are made by «System 1»: an automatic, rapid mode of thinking that operates without conscious control and is prone to mental shortcuts known as heuristics. Whilst this mechanism is essential to our day-to-day survival, it also proves to be the source of numerous errors in operational judgement. Conversely, «System 2» – which is slow and deliberate – lies at the very heart of critical thinking but accounts for only a minority of our conscious decisions.
This raises a fundamental question: as a manager, is your decision-making process rigorous enough to avoid these hidden pitfalls and harness collective intelligence without succumbing to the illusions of technology?
Critical thinking in management: what are we talking about?
Critical thinking: what is it? It is a reflective process aimed at assessing the validity of information or reasoning before making a decision. It enables us to move beyond the «we’ve always done it this way» mindset, to examine a situation from multiple angles and choose the best course of action.
[Key figures] The cost of poor decision-making in business
Bad decisions come at a cost that far exceeds that of a one-off mistake. In a McKinsey survey (Flaws in strategic decision making: McKinsey Global Survey results, 2009) of 2,207 executives, only 28 per cent felt that their company’s strategic decisions were generally good. And 60 per cent felt that bad decisions were about as common as good ones.
The cost is also linked to the time spent on inefficient decision-making processes. Another McKinsey survey (‘Decision making in the age of urgency’, 2019), conducted amongst more than 1,200 managers, estimates that, in a typical Fortune 500 company, respondents reported spending an average of 37 % of their time on decision-making, more than half of which was deemed to be used inefficiently.
Finally, the quality of the process has a direct impact on performance. Out of 1,048 decisions analysed by McKinsey (The Case for Behavioural Strategy, 2010), the quality of the process was a stronger predictor of results (+7 % in ROI) than the volume of analysis produced.
In other words: the cost of a bad decision lies not only in its outcome, but also in the time wasted in making it, getting it approved and rectifying its consequences.
Source : McKinsey Quarterly – «The case for behavioural strategy» by Dan Lovallo and Olivier Sibony.
Identifying and overcoming cognitive biases in management
Cognitive biases are distortions in the logical processing of information. They are also referred to as mental shortcuts. These are often exacerbated by stress or a sense of urgency. In situations of high time pressure or stress, our attention may focus on a more limited amount of information and our ability to consider several options may diminish. The amygdala, which is particularly involved in detecting threats and emotional responses, may become more reactive. At the same time, certain control processes—particularly those involving the prefrontal cortex—may become less effective, favouring faster, habitual or intuitive responses. This pressure can therefore reinforce certain mental shortcuts and cognitive biases.
To deal with this, the modern manager must learn to identify the three common biases which undermine its objectivity on a day-to-day basis.
The anchoring bias
An unintentional tendency to focus disproportionately on the first impression or the first set of figures received, thereby precluding a balanced, comprehensive analysis.
Confirmation bias
A reflexive tendency to focus solely on information that supports our preconceived assumptions or beliefs, whilst actively ignoring evidence to the contrary.
The conformity bias
A social mechanism that encourages an individual to fall in line with the presumed opinion of the majority in order to fit in or avoid conflict, leading to the harmful phenomenon of groupthink.
How can cognitive biases be counteracted in decision-making?
First of all, take a step back to ease the pressure…
Above all, put in place rigorous collective processes.
decision checklist inspired by the work of the researcher Olivier Sibony, proposes 5 essential checkpoints before finalising any major decision.
- The benefit : Do the people recommending a solution have a particular interest in seeing it adopted?
- The halo effect : Are we disproportionately influenced by the charisma, authority or past successes of the person who came up with the idea, rather than by the factual relevance of its content?
- The alternative : Have we seriously considered at least two other credible options, or have we put together our case simply to justify a single preferred option?
- The source : Do the data presented stem from verifiable raw facts or from interpretations that are already biased?
- The disagreement : Has anyone put forward a well-reasoned opposing view? If not, please volunteer a colleague to play the devil’s advocate and look for weaknesses.
[The 3 keys to decision-making]
Make only those decisions that you alone must make.
If the decision is straightforward, involves little at stake or can be reversed, delegate… or make a quick decision!
Don’t be ashamed to change your mind.
You can only know whether a decision is «right» if…’in hindsight, not in the moment. The mistake is therefore part of the decision. The only real problem, then, is to stick to one’s decision.
Set yourself a deadline for making a decision even if you don’t have all the information.
Critical thinking and AI: how can we maintain our discernment?
The widespread integration of artificial intelligence tools into managerial practices is creating a new form of decision-making illusion. We were already familiar with hallucinations, source illusions and AI training biases, such as gender bias. An MIT study published in February 2026 highlights a worrying phenomenon known as «chatbot complacency» (excessive and self-serving flattery). Language models have an unfortunate tendency to flatter the user and artificially reinforce the direction of their queries rather than restoring the raw truth.
[Case study] Using critical thinking to analyse an AI-generated action plan for difficult feedback
The plot
A manager needs to hold a performance review with a technical expert who is essential to the project, but whose behaviour is unsettling the junior staff (public criticism, withholding information). Fearing that the expert might resign, the manager asks an AI to draft the talking points for the meeting.
The machine’s proposal
The AI instantly generates a plan centred on «absolute kindness», suggesting that one should first spend a considerable amount of time highlighting their unique operational skills before addressing any relationship issues.
The manager’s critical review
By applying critical thinking, the manager identifies two blind spots in the algorithmic proposal.
- AI presupposes perfect rationality. It overlooks the possibility that the expert might be in a state of complete denial or actively on the offensive, using their strategic importance as a means of blackmail during the exchange.
- AI lacks courageous neutrality. To please the manager, the system suggests a «half-hearted» plan that avoids setting firm limits.
Management practice
The manager chooses to amend the plan by incorporating the actual human context. He strips away the flattery and adds a clear framework based on the exact wording of the comments and the specific consequences should the situation persist. He may also ask the AI to examine the blind spots in the proposal, particularly regarding the expert’s stance and how to prepare for different scenarios.
Whilst AI’s contributions are relevant, its responses may reproduce or amplify certain biases. Generative models rely on statistical patterns learnt from their training data and context. As a result, they may reproduce dominant representations, lack context or perspective, and, at times, reinforce the user’s beliefs rather than challenging them.
[Also read] Best practices for working as a team with AI
This inevitably presents a challenge to managers in terms of how they process information, balancing intuitive analysis with critical analysis.
Intuitive analysis versus critical analysis
| Assessment criteria | Intuitive analysis (System 1 / Machine) | Critical analysis (System 2 / Conscious Manager) |
| Cockpit | Passive stance: immediate acceptance of the deliverable. | Active stance: methodical scepticism and suspension of judgement. |
| AI processing | Delegating the decision or placing blind trust in the text. | Treating the response as mere raw material. |
| Methodological rigour | Focus on short-term time savings. | Analysis of blind spots and identification of weaknesses. |
Collective intelligence: how can we avoid group bias?
Collectives are a powerful tool for solving complex problems. However, they also provide fertile ground for group bias. The American journalist James Surowiecki has shown that, for a collective decision to be sound, it must meet four fundamental conditions:
- the diversity of opinions
- the independence of judgements
- the decentralisation of knowledge
- a clear mechanism for aggregating opinions
Without this framework, the group can easily fall prey to biases. The most common of these is groupthink. This bias stems from our need for acceptance and conformity to social norms. Out of fear of rejection, a person may rationalise their adherence to a group by citing the group’s intentions – even if they do not agree with the group’s values.
4 ways to develop critical thinking within a team
To safeguard everyone’s freedom of thought and strengthen collective intelligence in the long term, managers have a range of rituals and structured methods that they can incorporate into their day-to-day management.
- Le brainwriting : Ask each member of staff to write down their ideas and objections individually on a Post-it note before any verbal discussion takes place. This prevents the first person to speak from exerting an overwhelming and immediate influence (anchor bias).
- The leader speaks last : adopt an approach whereby the manager always speaks last. This helps to ensure that the team’s opinions are not influenced by the weight of hierarchical authority.
- The Contradiction in the Making : appoint a different person at each meeting whose explicit and valued role is to organise and prepare a counter-argument in the run-up to the decision-making meeting. This naturally requires that they have been provided with all the relevant information prior to the meeting.
- The pre-mortem.
[The pre-mortem method]
Unlike a post-mortem (carried out after a failure), the pre-mortem is carried out before to launch a major project.
3 steps
1/ Imaginary projection
Gather the team together and say:
«Let’s imagine it’s six months or a year from now. The project is a complete disaster. It has failed on every front.»
2/ The individual self-assessment (5 minutes)
Each participant silently notes down all the specific reasons behind this tragic failure.
For example: weak signals being overlooked, technical risks, a lack of resources.
3/ The accumulation of faults and preventive action
First, pool the identified risks. Then, turn these potential causes of failure into a preventative action plan to safeguard the decision before it is implemented.
How can we prevent critical thinking from becoming an obstacle to decision-making?
Critical thinking is an essential safeguard. That is why many training and education programmes seek to develop this capacity for discernment. However, this must not lead to paralysis. Managers must therefore be wary of 3 major pitfalls.
1/ Decision-making fatigue
Our capacity for critical thinking is a cognitively demanding skill that becomes depleted as the day goes on. Make high-stakes decisions when teams are focused and fully engaged in the decision-making process.
2/ Analysis paralysis
Wanting to «know everything» and endlessly accumulating data before making a decision means missing out on strategic opportunities.
3/ The social cost
Unchecked, excessive doubt can be perceived by the team as mistrust, disengagement or negativity.
Making better decisions without over-analysing: the 40/70 rule
The so-called 40/70 rule is attributed to General Colin Powell. It suggests that one should not make a decision too early, based on very little information, but nor should one wait for a certainty that is unattainable.
To find out exactly when to stop analysing and take action, apply the 40/70 rule.
- Less than 40 % of information : You do not have enough factual evidence. Making a decision at this stage would be a high-risk gamble.
- Over 70 % of information : Seeking further information leads to diminishing returns. You fall into analysis paralysis and the opportunity slips away.
- The decision-making area : this is where the real managerial trade-off lies. The’managerial intuition (based on practical experience) bridges the gap between 70 % and 100 %, enabling a confident decision.
These percentages should be understood as a managerial metaphor, not as scientific thresholds. Indeed, in most situations, no one knows how to quantify exactly what «40 %» or «70 %» of the necessary information represents.
In conclusion, critical thinking is a practical skill that is honed on a daily basis. It does not mean doubting everything, but rather questioning more effectively the factors that guide our decisions: our intuitions, the views of the group, the available data… or the responses from an AI. By taking time to step back, testing assumptions and encouraging debate, managers can improve the quality of their judgements without slowing down the process. In the age of AI, their role remains essential: to use tools to inform their decisions, without delegating their judgement to them.
