Why AI Could Weaken Corporate Judgment

Artificial intelligence was meant to save us time. It sometimes seems to produce the opposite effect.

Generative tools enable us to summarize faster, write more, multiply analyses, and accelerate exchanges. Yet the time saved is not always transformed into space for reflection. It is often absorbed by new tasks, higher targets, and an intensification of digital work.

This paradox forms the starting point of an article by David Brooks published in The Atlantic under the title The People Who Will Thrive in the AI Age. The American journalist defends a powerful idea: when intelligence becomes abundant, the willingness to make an intellectual effort gains value.

His analysis does not seek to pit AI users against those who would refuse technology. Rather, it questions how each person chooses to use it. Some will progressively delegate thinking to the machine. Others will use AI to learn, challenge their own ideas, and undertake more ambitious tasks.

This distinction goes far beyond personal productivity. It raises a deeper question: what becomes of the human being when a machine can read, write, search, and reason with them, or even in their place?

Why David Brooks’s analysis deserves to be read

David Brooks starts from a now familiar reality. Generative models considerably reduce the cost of many intellectual tasks. They produce a synthesis, structure reasoning, propose arguments, or search for references in seconds.

This availability transforms our relationship with difficulty.

When a convincing answer is immediately accessible, why continue alone with hesitant thinking? Why write an imperfect first draft? Why search at length for a source, compare several interpretations, or face the discomfort of a still poorly defined problem?

Brooks does not condemn this ease. He recognizes the benefits of AI and uses it himself. But he observes that part of our intellectual development depends precisely on the stages that technology now allows us to bypass.

Writing forces us to organize our thinking. Reading several authors allows us to distinguish nuances. Resisting a first answer develops judgment. Making mistakes, revisiting a hypothesis, and reformulating a problem all contribute to learning.

Not all difficulties obviously deserve to be preserved. Spending several hours reformatting a document or correcting syntax does not necessarily develop a distinctive skill. But some frictions are formative. Their systematic elimination can improve immediate production while progressively reducing autonomy.

This is why Brooks distinguishes, in essence, several attitudes toward AI.

Some users primarily seek to reduce effort. The tool becomes a means of obtaining a text, analysis, or answer more quickly, without going through the stages of reflection.

Others perceive the risk but give in to ease under the effect of professional pressure and lack of time.

A third category uses AI to extend their capabilities. These users do not merely ask it to solve a problem. They use it to test a hypothesis, identify a contradiction, explore literature, or examine a subject from several angles.

The difference therefore does not rest solely on technical mastery of the tool. It depends on the relationship each person maintains with intellectual effort.

Write first, confront later

The article’s interest also lies in its personal nature. David Brooks does not merely formulate a thesis. He describes the practices he tries to adopt to avoid excessive delegation of his thinking.

Before consulting a model, he begins by writing his own analysis. He produces an initial conclusion, however imperfect. He then asks the AI to challenge his reasoning, reveal its weaknesses, or propose perspectives he had not considered.

The sequence is important.

When AI intervenes before the formulation of personal thought, it can define the problem framework, select arguments, and influence the conclusion. The user then works within an intellectual architecture they have not built.

When it intervenes after initial reflection, it can become an instrument of contradiction. It no longer replaces reasoning. It tests it.

Brooks also explains that he seeks to alternate tasks performed with AI and those he carries out without assistance. The objective is not to artificially maintain old methods, but to continue exercising his writing, reading, and creative abilities.

This discipline reflects a simple idea: cognitive skills are not permanent acquisitions. They develop through use and can weaken when no longer solicited.

The librarian rather than the oracle

One of David Brooks’s most fruitful formulas consists of treating AI as a brilliant librarian rather than an oracle.

The oracle provides a supposedly definitive answer. The librarian opens a field of exploration. They indicate which authors have addressed a subject, what controversies exist, which schools of thought oppose each other, and which references can enrich reflection.

This difference in posture modifies the questions asked of the tool.

Instead of asking: “What should I think?” the user can ask:

• “Which authors have worked on this question?”
• “What hypotheses could contradict my analysis?”
• “What elements would invalidate my conclusion?”
• “What competing interpretations should I examine?”

AI then becomes an intellectual resource, but it does not receive the authority to conclude in place of the user.

This approach is particularly relevant in an environment where models produce fluent responses, apparently coherent, and sometimes difficult to distinguish from expert analysis. The quality of formulation can create an illusion of solidity. Yet a convincing answer is not necessarily correct, complete, or adapted to the context.

The user’s role therefore consists less in obtaining a production than in evaluating its relevance.

A possible cognitive polarization

Brooks’s most concerning thesis concerns the gap that could widen between users.

Some will mobilize AI to access knowledge more quickly, deepen their reflection, and undertake projects previously beyond their reach. The tool will increase their capacity for action.

Others will progressively delegate formulation, analysis, and research. They will be able to produce more, but risk understanding less deeply what they produce.

This polarization will not necessarily follow traditional divisions of degree, function, or level of expertise. It could depend more on curiosity, perseverance, the ability to tolerate uncertainty, and the willingness to continue learning when the tool makes the economy of thought possible.

An almost philosophical question then emerges: is the human being distinguished by their intelligence or by the desire they have to exercise it?

AI provides growing cognitive power. It does not choose the objectives worth pursuing, the problems that must be solved, nor the consequences we are prepared to assume.

For decades, companies have invested in their information systems. They are now investing in artificial intelligence systems. Tomorrow, their true competitive advantage may depend less on these technologies than on their ability to preserve and develop the collective judgment of their teams.

An organization capable of quickly producing an answer without being able to explain its reasoning gains operational efficiency but progressively loses strategic control.

From individual to organization

This reflection therefore goes beyond personal development. It raises a question still rarely addressed in companies: what becomes of an organization when its employees progressively delegate part of their intellectual effort?

A company is not merely an addition of individual skills. It relies on collective mechanisms allowing interpretation of a situation, confrontation of viewpoints, transmission of experience, and decision-making.

AI can strengthen these mechanisms. It can facilitate access to information, accelerate scenario preparation, and help teams explore a problem from several angles.

But it can also make them more opaque.

An analysis is generated, integrated into a presentation, taken up in a recommendation, then used to guide a decision. At each stage, the result circulates, but the initial assumptions become less visible.

The organization retains the conclusion. It progressively loses track of the reasoning.

The risk is therefore not that employees use AI. It is that the company ceases to understand how its own decisions are constructed.

Intellectual effort does not disappear, it changes location

Artificial intelligence does not eliminate intellectual effort. It shifts its value.

It becomes less useful to spend several hours on document formatting, producing an initial synthesis, or correcting a first version.

On the other hand, problem formulation, source verification, identification of blind spots, understanding context, and arbitration between several options take on greater importance.

Imagine two companies that deploy exactly the same AI assistant to their sales teams.

In the first, salespeople use the tool to write their proposals more quickly. Deadlines decrease, document volume increases, and targets are raised. AI accelerates production.

In the second, the tool also generates an initial proposal. But this version becomes the starting point for collective work. The team examines commercial assumptions, anticipates client objections, verifies data, and evaluates contractual risks. The manager asks salespeople to explain why certain options were chosen.

The first company produces more. The second develops stronger commercial capability. Both organizations use the same technology. They do not reinvest the time and attention freed up in the same way.

It is in this difference that true value creation lies.

Management facing production that has become easy

This transformation directly modifies the manager’s role.

When writing a report or preparing an analysis required several hours of work, the final result gave certain indications about its author’s involvement and understanding.

With AI, this relationship becomes less legible.

A highly elaborate document can be produced in minutes. A convincing recommendation can reproduce reasoning its author does not master. Professional formulation can mask insufficient verification.

The manager can therefore no longer merely ask:

“Is the work finished?”

They must also ask:

“How did you arrive at this conclusion?”

This question makes visible the sources used, assumptions made, options discarded, controls performed, and uncertainties that remain.

The role of management is not to monitor every interaction with AI. It consists of establishing quality requirements in a context where production becomes easier, but where understanding can become more fragile.

A well-written answer should not be confused with solid analysis. A quick recommendation must be able to be explained, contested, and defended.

What the company rewards shapes usage

The relationship with intellectual effort does not depend solely on individual willingness. It is also influenced by evaluation systems and objectives set by the organization. A company that primarily measures speed and volume produced will logically encourage the most automated uses. Teams will seek to process more cases, generate more content, and reduce deadlines.

An organization that values quality of reasoning, risk identification, originality of proposals, and the ability to justify a decision will encourage more demanding uses. When management only asks to go faster, AI serves speed.

When it asks to better understand, confront several scenarios, and improve decision quality, the tool can serve collective learning. AI governance is therefore not limited to security, compliance, or data protection. It also concerns the behaviors the company chooses to value.

This observation moreover joins a lesson that has returned recurrently during interviews conducted for EntrepreneurIA. The most advanced leaders in AI adoption do not describe a replacement of their thinking. They explain instead spending more time formulating problems, validating hypotheses, and making strategic arbitrations. For them, AI accelerates decision preparation. It does not decide in their place.

The questions leaders must now ask

Leaders do not have to choose between AI adoption and preservation of human capabilities. They must organize their articulation. Several practices can be implemented.

Any important recommendation should clearly distinguish facts, interpretations, and assumptions. Sources should be identified when the issue is financial, legal, strategic, or reputational. Teams should be invited to specify what could invalidate their conclusion.

Contradictory reviews can be organized to confront generated productions with other readings. Certain particularly sensitive decisions may also require initial reflection without assistance, followed by confrontation with AI.

Does the time freed by AI serve:

• to produce more or to better understand?
• to handle more requests or to strengthen client relationships?
• to multiply analyses or to improve their quality?
• to reduce costs or to develop skills?

A technology does not create value solely because it accelerates a task. Value depends on how the organization reinvests the capabilities it frees.

Preserving the will to think

David Brooks invites us to preserve our intellectual effort. His analysis is not a nostalgic defense of a world without artificial intelligence. It reminds us that a capacity that is no longer exercised can progressively weaken.

Companies will need to extend this reflection. They will need to build environments in which AI augments human capabilities without replacing the mechanisms that allow understanding, debating, and deciding collectively. The challenge is not to preserve all difficulties. Some tasks must be simplified, automated, or eliminated. But the elimination of useless effort must not lead to that of formative effort.

Intellectual effort is progressively leaving repetitive tasks to concentrate on what truly determines the quality of a decision: posing the problem, verifying assumptions, understanding context, interpreting consequences, arbitrating, and assuming responsibility.

In a world where abundant intelligence will be accessible to a growing number of individuals and companies, the advantage will no longer reside solely in the tools used. Tomorrow, companies will not be differentiated by the technologies they have access to, but by the quality of collective judgment they have managed to preserve. Artificial intelligence makes answers abundant. Judgment, however, will remain a skill to be built.