Why AI Makes Smart People Wrong Faster

For generations, intelligence was measured by the quality of our answers. Expertise became one of the defining characteristics of leadership, science, medicine, business—and virtually every profession built on knowledge.

Artificial intelligence is quietly changing that equation.

The greatest risk is no longer a lack of information. It is asking the wrong question with complete confidence—and receiving an excellent answer that makes the mistake even harder to recognize.

Ironically, the people most vulnerable may not be beginners. They may be the most experienced among us. Years of success help us recognize patterns, but they also make some assumptions so familiar that we stop seeing them.

Nearly forty years ago, Harvard professor Chris Argyris described this hidden phenomenon as skilled incompetence: our tendency to unconsciously protect the very assumptions that once made us successful.

Artificial intelligence did not create this human blind spot.
It may simply become the most powerful system ever built for quietly automating it.

Artificial intelligence is simply the catalyst that exposes a timeless human challenge. The defining question is no longer whether we can produce better answers. It is whether we are still asking the right questions.

The New Paradox

Artificial intelligence is transforming knowledge work at extraordinary speed. It retrieves information, analyzes complex data, generates persuasive arguments, and produces sophisticated solutions within seconds. Capabilities that once distinguished experts are becoming increasingly accessible to everyone.

At first glance, the conclusion seems obvious. If answers become faster, more complete, and more accurate, our decisions should naturally improve as well.

But artificial intelligence begins with the questions we ask. It does not decide whether those questions rest on assumptions that should first be challenged.

In other words, AI can dramatically improve the quality of an answer without ever examining the thinking that produced the question.

That is the paradox.

As answers become increasingly abundant, the ability to ask better questions may become one of the most valuable human capabilities of all.

When Intelligence Protects Its Own Assumptions

Nearly four decades before artificial intelligence entered everyday life, Harvard professor Chris Argyris described a remarkably human pattern that he called skilled incompetence.

His insight was both simple and unsettling.

As we gain experience, many of the mental models that help us succeed become so familiar that we stop noticing them. They no longer feel like assumptions. They simply become the way we see the world.

Every area of expertise gradually creates invisible assumptions.

Most of the time, these assumptions are enormously valuable. They allow us to recognize patterns, make decisions efficiently, and navigate complexity without having to rethink every problem from first principles.

But they can also quietly narrow our thinking.

The more often an assumption has served us well, the less likely we are to question it. Over time, what once accelerated our learning can begin to limit it without our realizing it.

Argyris observed that this process is rarely intentional. We do not consciously defend our assumptions. We simply continue building upon them because they have become automatic.

He called these patterns defensive routines because they unconsciously protect the mental models that once made us successful, even when those models deserve to be questioned.

Artificial intelligence introduces a new challenge.

It does not create these assumptions, and it does not question them. It accepts them as its starting point and quietly automates them.

That may be one of the most important—and least discussed—leadership challenges of the AI era.

When Thinking Becomes Automatic

Experience is one of our greatest advantages. Every experience becomes part of an internal library that helps us navigate future situations. Some experiences tell us what works. Others teach us what to avoid. Together, they allow us to recognize familiar patterns and respond more quickly and effectively without having to rethink every decision from first principles.

That is precisely what expertise is designed to do. It makes good judgment more efficient by allowing us to build on what we have already learned.

Over time, however, the patterns that once required careful thought gradually become automatic. They no longer feel like assumptions or interpretations. They simply become the way we see the problem.

Artificial intelligence changes what happens next.

Once those assumptions become embedded in our thinking, AI can make reasoning built upon them dramatically faster, broader, and more persuasive. It does not determine whether the underlying assumptions are correct. It simply extends the thinking it is given.

The answers become faster. The analyses become richer. The arguments become more persuasive. Naturally, our confidence grows.

Yet confidence is no longer a reliable indicator of good thinking. It may simply reflect better technology built upon assumptions that were never questioned.

The greatest risk is not that AI will give us poor answers. It is that we will rely on answers based on incorrect assumptions.

The Organizational Consequence

Organizations are shaped by people, but they are sustained by the ways those people learn to think together. Over time, the assumptions that guide experienced individuals gradually become embedded in everyday decisions, processes, meetings, incentives, and hiring practices. What begins as individual thinking slowly becomes organizational culture.

Culture is reflected not only in the answers an organization gives. It is equally reflected in the questions people feel comfortable asking.

Some questions are welcomed because they reinforce the way the organization already thinks. Others quietly disappear—not because they are explicitly discouraged, but because people learn that they create discomfort, challenge long-held assumptions, or rarely lead to action.

New employees quickly recognize these unwritten rules. Long before they fully understand the business, they begin to understand its culture. They learn which questions are appreciated, which are politely ignored, and which are better left unasked. Without realizing it, they gradually adapt their own thinking to the thinking of the organization.

Artificial intelligence introduces a new dimension.

It not only automates the invisible assumptions of individuals. It can also reinforce the assumptions embedded within an organization's culture. If the questions remain narrow, AI will produce increasingly sophisticated answers to those same questions. If the underlying assumptions deserve to be challenged, AI has no reason to challenge them. It simply extends the thinking it is given.

The leadership challenge therefore changes as well.

For generations, leaders were expected to provide better answers. Increasingly, their most important responsibility may be something different: creating an environment where better questions continue to be welcomed.

Organizations continue learning not because they possess the smartest people, but because they preserve the ability to question even their most successful assumptions.

Final Thoughts

The defining skill of the AI era will no longer be producing better answers.

It will be recognizing when a familiar question no longer deserves the answer we've always been trying to optimize.

The most valuable questions of the AI era are no longer those we ask technology. They are the ones we first ask ourselves.

What culture am I creating? Which assumptions guide my decisions? Which hypothesis have I stopped questioning? Which questions have I stopped asking myself?

Every breakthrough begins with a better question and often that question begins by challenging an old assumption.