The most dangerous answer is the one that sounds right

Imagine two senior executives receiving exactly the same recommendation. Their organisation is considering a significant change, and a team of specialists has produced a proposal explaining why the change is necessary, what benefits it will deliver and how it should be implemented.

The evidence appears convincing, the financial projections are encouraging and the people presenting the recommendation have considerable expertise.

The first executive thanks the team and approves the proposal. The second also recognises the quality of the presentation but asks several additional questions. What assumptions underpin the recommendation? What evidence might contradict the conclusion? What would need to happen for the proposal to fail?

The second executive is not necessarily more intelligent or experienced than the first. They may ultimately approve exactly the same proposal. The difference is that one executive has evaluated how convincing the information appears, while the other has examined how reliable it actually is. This distinction matters because some of the most dangerous information leaders encounter is not obviously incorrect. It is information that sounds entirely reasonable.

Many poor decisions begin with information that appears logical, is presented confidently and fits comfortably with what decision-makers already believe. The problem is that human beings are psychologically inclined to accept explanations that make sense, particularly when those explanations come from sources they regard as credible.

Why convincing information feels accurate

Our brains constantly seek patterns, connect information and develop explanations that help us understand the world. Because we cannot investigate everything, we rely on mental shortcuts, using familiarity, consistency and credibility to decide what can reasonably be accepted.

One consequence is that information which is easy to understand often feels more believable. Psychologists refer to this as processing fluency. When an explanation is clear and coherent, our brains process it with relatively little effort. That ease can influence our judgement of its credibility, even though a well-presented argument is not necessarily accurate.

Consider a manager explaining why employee turnover has increased. They might argue that younger employees are less loyal and more interested in changing jobs than previous generations. The explanation sounds familiar and provides a straightforward answer to a complicated problem. Yet the actual causes could involve poor management, limited career development or remuneration. By accepting the

explanation that sounds right, the leader may overlook the information needed to address the problem.

Confirmation bias makes this more difficult. We naturally give greater weight to information that supports what we already believe and scrutinise contradictory information more closely. A leader who believes a department is inefficient may readily accept evidence supporting that view while questioning evidence suggesting it is performing well. Information stops being evaluated according to its quality and starts being evaluated according to how comfortably it fits an existing opinion.

Expertise is not the same as accuracy

As leaders become more senior, they increasingly depend on the knowledge of other people. A chief executive cannot personally possess the technical expertise of every accountant, engineer, lawyer or operational manager within an organisation. Effective leadership therefore requires trusting people who understand subjects better than we do.

However, respecting someone’s expertise is not the same as treating their conclusions as unquestionable. Expertise increases the likelihood that someone understands a subject, but it does not eliminate mistakes, incomplete information or flawed assumptions. Specialists can also become so familiar with their own area that they overlook alternative perspectives.

This creates a challenge for leaders who need to question people with greater technical knowledge. Some hesitate because they worry that asking questions will expose their lack of understanding. Others fear that challenging an expert will be interpreted as questioning their competence. Consequently, leaders sometimes approve recommendations they do not fully understand because the alternative feels uncomfortable.

Leadership does not require knowing more than the expert. It requires creating the conditions in which expertise can be examined. A leader may not understand every technical detail of a proposed system, but they can ask what assumptions have been made, what alternatives were considered and what evidence would cause the expert to reconsider their recommendation.

Challenge the information, not the person

People often interpret challenges to their ideas as challenges to themselves. When someone has invested considerable time developing a proposal, questioning its assumptions can feel like questioning their judgement. The more closely their professional identity is connected to their expertise, the more personal that challenge may become.

This is why the way leaders question information matters. There is a considerable difference between asking, “Why would you recommend something like that?” and asking, “Can you help me understand the assumptions behind this recommendation?” Both concern the same issue, but the first can imply incompetence while the second invites explanation.

Leaders can acknowledge someone’s expertise while explaining that important decisions require assumptions to be tested. Questions should focus on understanding the reasoning, identifying limitations and exploring alternatives rather than establishing who is right or wrong.

Leaders must also examine their own reactions when their ideas are questioned. If a manager becomes defensive whenever someone challenges their recommendation, employees learn that agreement is safer than critical thinking. Over time, the leader may become surrounded by people who present information designed to gain approval rather than expose weaknesses. The absence of disagreement can then be mistaken for evidence that everyone supports the decision.

A healthy leadership culture makes it possible to disagree without turning disagreement into conflict. The purpose of questioning is not to demonstrate superiority or win an argument. It is to improve the quality of the decision.

Artificial intelligence makes the problem more complicated

The rapid adoption of artificial intelligence introduces another dimension to this challenge. AI can produce detailed recommendations, persuasive explanations and sophisticated-looking analysis within seconds. It can organise complex information into clear arguments and present conclusions in language that sounds remarkably authoritative.

These capabilities offer significant benefits, but they also create a psychological risk. An answer may be well structured and supported by plausible explanations while containing factual inaccuracies, invented references or flawed reasoning. Depending on the system and task, AI may present a questionable conclusion with the same polished confidence as a correct one.

There is also the danger of automation bias, our tendency to place excessive trust in recommendations produced by automated systems. When technology produces an answer quickly and presents it professionally, people may become less inclined to question its reasoning. This is particularly concerning when AI recommendations influence recruitment, financial decisions or assessments of employee performance.

One useful exercise is to ask an AI system questions about a subject you know exceptionally well. Because you already understand the subject, you are better positioned to recognise factual mistakes, questionable assumptions and occasions when AI presents an oversimplified explanation with unwarranted confidence. You may be surprised by how convincing an incorrect answer can sound.

This exercise provides an indication of the system’s limitations and helps develop a more realistic level of trust. However, accuracy in one familiar subject does not guarantee accuracy in another. Reliability varies according to the task, the information available and the questions being asked. The purpose is not to establish that AI is either trustworthy or untrustworthy, but to understand why its answers require judgement.

The appropriate response is not to reject artificial intelligence. AI can be extremely useful for exploring options and supporting analysis. However, leaders need to distinguish between using AI to inform their judgement and allowing AI to replace it. For consequential decisions, important facts and assumptions should be independently verified.

The questions that improve decisions

One of the most useful habits leaders can develop is deliberately looking for information that challenges a recommendation they find convincing. This is harder than it sounds because once an explanation makes sense, our natural inclination is to stop searching. We feel that we have understood the problem and can move towards a decision.

Yet understanding an argument is not the same as testing it. A recommendation might be internally consistent while resting on an incorrect assumption.

Financial projections may be mathematically accurate but depend on unrealistic expectations about customer behaviour. A proposed restructure may address a visible symptom without resolving the underlying cause.

Leaders can improve their decisions by asking what would need to be true for a recommendation to succeed and what circumstances might cause it to fail.

They should explore whether alternative explanations have been considered and whether the evidence genuinely supports the conclusion. These questions are particularly valuable when leaders personally favour the proposed course of action.

This is not an argument for endless analysis or challenging every minor decision. The amount of scrutiny should reflect the significance of the decision. A routine operational choice does not require the same examination as a major investment or organisational restructure. The objective is intellectual discipline, not paralysis through constant questioning.

The discipline to remain curious

Perhaps the greatest danger facing senior leaders is not that they become less intelligent as their careers progress, but that their success makes their assumptions less likely to be challenged.

People become increasingly reluctant to question those

Confirmation bias makes this more difficult. We naturally give greater weight to information that supports what we already believe and scrutinise contradictory information more closely.

A leader who believes a department is inefficient may readily accept evidence supporting that view while questioning evidence suggesting it is performing well. Information stops being evaluated according to its quality and starts being evaluated according to how comfortably it fits an existing opinion.

Expertise is not the same as accuracy

As leaders become more senior, they increasingly depend on the knowledge of other people. A chief executive cannot personally possess the technical expertise of every accountant, engineer, lawyer or operational manager within an organisation. Effective leadership therefore requires trusting people who understand subjects better than we do.

However, respecting someone’s expertise is not the same as treating their conclusions as unquestionable. Expertise increases the likelihood that someone understands a subject, but it does not eliminate mistakes, incomplete information or flawed assumptions. Specialists can also become so familiar with their own area that they overlook alternative perspectives.

This creates a challenge for leaders who need to question people with greater technical knowledge. Some hesitate because they worry that asking questions will expose their lack of understanding.

Others fear that challenging an expert will be interpreted as questioning their competence. Consequently, leaders sometimes approve recommendations they do not fully understand because the alternative feels uncomfortable.

Leadership does not require knowing more than the expert. It requires creating the conditions in which expertise can be examined.

A leader may not understand every technical detail of a proposed system, but they can ask what assumptions have been made, what alternatives were considered and what evidence would cause the expert to reconsider their recommendation.

Challenge the information, not the person

People often interpret challenges to their ideas as challenges to themselves. When someone has invested considerable time developing a proposal, questioning its assumptions can feel like questioning their judgement. The more closely their professional identity is connected to their expertise, the more personal that challenge may become.

This is why the way leaders question information matters. There is a considerable difference between asking, “Why would you recommend something like that?” and asking, “Can you help me understand the assumptions behind this recommendation?” Both concern the same issue, but the first can imply incompetence while the second invites explanation.

Leaders can acknowledge someone’s expertise while explaining that important decisions require assumptions to be tested. Questions should focus on understanding the reasoning, identifying limitations and exploring alternatives rather than establishing who is right or wrong.

Leaders must also examine their own reactions when their ideas are questioned. If a manager becomes defensive whenever someone challenges their recommendation, employees learn that agreement is safer than critical thinking. Over time, the leader may become surrounded by people who present information designed to gain approval rather than expose weaknesses. The absence of disagreement can then be mistaken for evidence that everyone supports the decision.

A healthy leadership culture makes it possible to disagree without turning disagreement into conflict. The purpose of questioning is not to demonstrate superiority or win an argument. It is to improve the quality of the decision.

Artificial intelligence makes the problem more complicated

The rapid adoption of artificial intelligence introduces another dimension to this challenge. AI can produce detailed recommendations, persuasive explanations and sophisticated-looking analysis within seconds. It can organise complex information into clear arguments and present conclusions in language that sounds remarkably authoritative.

These capabilities offer significant benefits, but they also create a psychological risk. An answer may be well structured and supported by plausible explanations while containing factual inaccuracies, invented references or flawed reasoning. Depending on the system and task, AI may present a questionable conclusion with the same polished confidence as a correct one.

There is also the danger of automation bias, our tendency to place excessive trust in recommendations produced by automated systems. When technology produces an answer quickly and presents it professionally, people may become less inclined to question its reasoning.

This is particularly concerning when AI recommendations influence recruitment, financial decisions or assessments of employee performance.

One useful exercise is to ask an AI system questions about a subject you know exceptionally well. Because you already understand the subject, you are better positioned to recognise factual mistakes, questionable assumptions and occasions

when AI presents an oversimplified explanation with unwarranted confidence. You may be surprised by how convincing an incorrect answer can sound.

This exercise provides an indication of the system’s limitations and helps develop a more realistic level of trust. However, accuracy in one familiar subject does not guarantee accuracy in another.

Reliability varies according to the task, the information available and the questions being asked. The purpose is not to establish that AI is either trustworthy or untrustworthy, but to understand why its answers require judgement.

The appropriate response is not to reject artificial intelligence. AI can be extremely useful for exploring options and supporting analysis. However, leaders need to distinguish between using AI to inform their judgement and allowing AI to replace it. For consequential decisions, important facts and assumptions should be independently verified.

The questions that improve decisions

One of the most useful habits leaders can develop is deliberately looking for information that challenges a recommendation they find convincing. This is harder than it sounds because once an explanation makes sense, our natural inclination is to stop searching. We feel that we have understood the problem and can move towards a decision.

Yet understanding an argument is not the same as testing it. A recommendation might be internally consistent while resting on an incorrect assumption.

Financial projections may be mathematically accurate but depend on unrealistic expectations about customer behaviour. A proposed restructure may address a visible symptom without resolving the underlying cause.

Leaders can improve their decisions by asking what would need to be true for a recommendation to succeed and what circumstances might cause it to fail.

They should explore whether alternative explanations have been considered and whether the evidence genuinely supports the conclusion. These questions are particularly valuable when leaders personally favour the proposed course of action.

This is not an argument for endless analysis or challenging every minor decision. The amount of scrutiny should reflect the significance of the decision. A routine operational choice does not require the same examination as a major investment or organisational restructure. The objective is intellectual discipline, not paralysis through constant questioning.

The discipline to remain curious

Perhaps the greatest danger facing senior leaders is not that they become less intelligent as their careers progress, but that their success makes their assumptions less likely to be challenged. People become increasingly reluctant to question those

with authority, while leaders themselves may grow more confident in the judgement that helped them achieve their position.

This creates an environment in which convincing information can travel upwards without being adequately tested.

Employees may avoid presenting contradictory evidence, specialists may assume their expertise should be sufficient and executives may mistake a polished presentation for a sound recommendation.

Artificial intelligence can intensify this problem by producing persuasive analysis faster than people can critically evaluate it.

Effective leaders recognise that confidence, expertise and accuracy are three different things. They respect expertise without surrendering judgement, welcome challenges without interpreting them as personal criticism and remain willing to question explanations that support their existing beliefs. Their curiosity is not a sign of uncertainty about their authority, but an acknowledgement that authority does not guarantee understanding.

Ultimately, the quality of leadership decisions depends not simply on the information available, but on how that information is evaluated.

A recommendation should not be accepted merely because it comes from an expert, supports an existing belief or has been produced by sophisticated technology. Leaders need the discipline to examine what lies beneath the confidence of the presentation, particularly when the answer appears obvious.

After all, the information most likely to mislead us is often not the information that sounds wrong, but the information that sounds so right we never think to question it.

(Mark Wager is one of the worlds top Leadership experts. Mark is due to visit Fiji later this year, if you would like Mark to deliver training at your company you can contact him directly at Mark@ali.org.nz)