The First Victim of AI … Poor Leadership
“Where is the wisdom we have lost in knowledge? Where is the knowledge we have lost in information?”
- T. S. Eliot
Artificial intelligence (AI) is becoming one of the most powerful tools ever placed in the hands of a leader. It can research a problem, analyze enormous amounts of information, compare ideas, challenge assumptions, summarize meetings, build presentations, evaluate trends, and produce possible solutions in a fraction of the time those jobs once required. The technology will continue to improve, and leaders who refuse to learn to use it will eventually become dinosaurs, putting themselves and their organizations at a disadvantage.
But access to more information has never guaranteed better leadership.
That may become one of the great misunderstandings of the AI revolution. We are beginning to confuse the ability to obtain an answer with the ability to understand what that answer means. Those are very different skills. AI can dramatically shorten the distance between a question and information, but there is still another distance to travel between information and a good decision. That distance is where leadership lives.
The numbers already show how quickly AI is moving into the workplace. Gallup reported in 2026 that 30 percent of U.S. employees were using AI at work several times a week or more, while 47 percent said their organizations had integrated AI technology into their operations. Yet Gallup found something even more important than the adoption numbers. Employees whose managers actively supported the use of AI were far more likely to use it frequently and substantially more likely to believe it improved how work was being done. The technology mattered, but the leader responsible for helping people use it mattered even more.
AI is not removing leadership from the equation. It is putting leadership under a brighter light.
Information Has Never Been the Same as Judgment
For most of history, information was difficult to acquire. Leaders who possessed better information frequently possessed an enormous advantage. Generals needed scouts. Businesses depended on market research. Coaches traded film and traveled to clinics because discovering something another program did not know could create a competitive advantage.
That world has changed.
Today, a young coach can sit at a computer and access training information that took me years to accumulate. A small-business owner can research markets that once required a team of analysts. A manager can ask AI to compare several possible strategies and receive a detailed response before everyone else has returned from lunch. As AI continues to improve, the information gap between people will narrow. The judgment gap will not.
I watched an earlier version of this happen when sports science began taking a much larger role in professional football. Coaches were suddenly inundated with numbers. GPS systems measured distance, speed, acceleration and workload. Force plates measured different aspects of strength and power. Recovery systems attempted to tell us how prepared an athlete was to work. Before long, numbers that had originally been introduced to provide coaches with additional information were beginning to influence how much an athlete should practice and, in some cases, whether he should practice at all.
My concern was never that we were collecting too much information. My concern was whether we understood enough about the information we were collecting to give it that much authority.
I kept asking questions. Where did these workload limits come from? Who established what constituted a red line? What athletes were used to create the standards? How much NFL-specific information existed when those recommendations were being made? If much of the early research had been developed through sports such as soccer, rugby and Australian football, how confidently could we apply those numbers to an NFL wide receiver, offensive lineman or linebacker whose physical demands were dramatically different? Those were not arguments against science. They were the questions science should demand.
The problem becomes even greater when averages begin replacing individuals. Two athletes may both have “WR” written beside their names on the roster, but that does not make them physically interchangeable. One may have spent years developing an enormous work capacity while another has not. One may play primarily on offense while another is taking offensive repetitions, special-teams snaps and scout-team work. One may weigh 185 pounds and win with speed while another weighs 225 pounds and spends part of every Sunday blocking linebackers and safeties. Putting both athletes underneath the same workload ceiling simply because they play the same position can become the analytical equivalent of saying an apple and a cheeseburger are both food, so nutritionally they should be treated the same.
What bothered me even more was when the limitations behind those numbers were not part of the conversation. Coaches did not need another person walking into the room and telling them, “The number says he is red, so he needs to practice less.” We needed someone capable of explaining why he was red, what evidence established that threshold, what weaknesses existed in the model, and whether the circumstances surrounding that particular athlete might make the number misleading. Information should have started the conversation, not ended it.
That experience is one reason I look at artificial intelligence with both excitement and caution. AI is going to give leaders access to more information than we have ever possessed. But the sophistication of the technology cannot become a substitute for understanding the source, limitations, and context of the information it provides. The more impressive the technology becomes, the easier it may be to stop asking the most important question: How do we know this is right?
The problem began when measuring an athlete started being confused with understanding an athlete.
A computer could tell us that a receiver had crossed a predetermined workload threshold. It could not automatically understand why he crossed it. Maybe injuries at his position forced him to take his normal special-teams work, scout-team repetitions, and additional repetitions with the starting offense. Maybe he had spent years developing a greater work capacity than another player at the same position. Maybe the number represented danger. Maybe it represented adaptation. The information was valuable, but someone still had to understand the football player standing in front of us. The same challenge is coming to every profession through AI.
A leader can ask AI to tell them which employee should be promoted, which department should be reduced, which product should be discontinued, or which strategy offers the greatest probability of success. The answer may contain outstanding information. It may identify patterns the leader overlooked. It may even expose weaknesses in the leader's original thinking. But the leader still has to know enough to recognize whether the answer makes sense.
That is why McKinsey argued earlier this year that AI makes human leadership more important, not less. AI can perform many tasks at extraordinary speed, but it cannot assume the human responsibilities of setting mindsets, building trust, holding people accountable, and making difficult organizational decisions. Their recommendation is essentially that leaders learn to use AI to think with them rather than allowing it to think for them.
That distinction may determine who becomes better because of AI and who becomes dependent upon it.
Better Answers Begin With Better Questions
One of the most overlooked leadership skills is the ability to ask the right question.
Early in my coaching career, I thought knowledge meant having answers. Experience eventually taught me that knowledge often means recognizing which questions still need to be asked. When something did not make sense in a training program, I learned not to immediately accept what I was told just because it came from someone with greater expertise in that area.
Where did this number come from? Who was studied? How large was the sample? Does this apply to football? Does it apply to this position? More importantly, does it apply to this athlete?
Those questions did not reject science. They made the science more useful. I often found that the person selling the snake oil had no real idea why; it was just what they were told. I knew when they were making stuff up when they would respond, “It’s too difficult to explain it.” This is where my background in history came out when I remembered an Albert Einstein quote, “If you can't explain it to a six-year-old, you don't understand it yourself.” This was the case with the guys who called themselves “sports scientists”.
AI will require the same discipline. A leader who asks a shallow question may receive a beautifully written shallow answer. A leader who begins with a faulty assumption may receive an incredibly sophisticated response built around that faulty assumption. The presentation of the answer can create a dangerous illusion because confidence and accuracy are not the same thing.
Air Canada learned that lesson in a very public way. A customer used the airline's chatbot while trying to understand its bereavement-fare policy. The chatbot supplied incorrect information about how the policy worked. When the customer later sought the refund he believed he had been promised, Air Canada argued that it should not be responsible for information supplied by the chatbot. A Canadian tribunal rejected that argument and held the airline responsible for the information its system provided.
Think about what happened there from a leadership perspective. The technology supplied the information, but the organization still owned the consequences.
That principle becomes much more important as AI begins participating in larger decisions. A leader cannot approve a strategy, watch it fail, and then explain that AI recommended it. You can delegate research. You can delegate analysis. You can ask technology to challenge your assumptions and provide alternatives. What you cannot delegate is ownership.
John F. Kennedy faced a dramatically different version of the information problem during the Cuban Missile Crisis in 1962. American intelligence had photographic evidence that Soviet nuclear missiles were being installed in Cuba. Kennedy had military advisers, intelligence officials, diplomats, and members of his Executive Committee presenting information and recommending different responses. Some supported military action. Others favored a naval blockade and additional diplomacy.
Kennedy's responsibility was not simply to collect the greatest amount of information. His responsibility was to determine which information mattered, understand what might happen after each possible decision, continue asking questions, and ultimately make a choice when nobody could guarantee the outcome.
Technology changes. Leadership responsibility does not.
AI Will Make Strong Leaders Better
There is another side of this discussion that cannot be ignored. AI should make good leaders significantly better.
A secure leader should love a tool capable of challenging his thinking. He can ask AI to find weaknesses in his plan, build the strongest argument against his position, identify information he may have overlooked, compare his strategy with historical examples, or show him how someone from another discipline might view the same problem. That is an extraordinary leadership advantage.
The insecure leader may use the same technology very differently. Instead of challenging what he believes, he can use AI to validate it. He can keep changing the question until the answer supports the decision he already wanted to make. Instead of becoming more curious, he becomes more certain. The technology has not corrected the weakness in his leadership. It has magnified it. Perfect case of Garbage In / Garbage Out. This is why we all need to stay focused on finding the right answer that is specific to our unique situations.
Sports analytics provide a useful comparison. Modern football coaches possess information previous generations never had. They can know the probability of converting fourth-and-two from a particular area of the field, how an opponent behaves against certain personnel groups, which coverage is most likely in a specific down-and-distance situation, and how game conditions alter expected outcomes. That information can improve the decision. It cannot make the decision.
The coach still knows whether his offensive line is controlling the line of scrimmage, whether his quarterback is seeing the field clearly, whether a particular player is injured, how the opponent has adjusted, and whether the situation in front of him resembles the thousands of situations used to build the probability. Analytics should make the coach smarter. They should never allow him to stop coaching. AI should serve leadership the same way.
This is why some of the most interesting AI research in 2026 is beginning to focus less on whether organizations possess AI and more on the managers responsible for implementing it. Harvard Business Review recently argued that middle managers may ultimately make or break AI adoption because they are the people translating executive enthusiasm into everyday behavior. Other HBR research found those same managers increasingly responsible for validating AI output, identifying mistakes, and teaching employees how to use the technology while still being expected to perform their normal responsibilities.
The organization can buy the technology. Leadership determines what happens next.
AI Cannot Give You Courage
There is another part of leadership that will become more valuable as information becomes easier to obtain. Courage. Leaders rarely struggle because absolutely no information exists. They struggle because the available information points in different directions. One report recommends expanding. Another recommends waiting. One expert sees opportunity. Another sees danger. Eventually somebody has to decide. AI will make those choices more informed, but it will not make them disappear.
Jeff Bezos has written about the importance of making many decisions before possessing all the information a leader would ideally like to have. His reasoning was that waiting for nearly complete certainty often means moving too slowly. That principle becomes even more relevant in an AI environment because leaders can convince themselves that one more search, another analysis, another simulation, or another prompt will finally eliminate uncertainty. It won't. There will always be another question to be asked.
At some point, gathering information becomes hiding behind information. The leader who continually asks AI for another answer may look analytical when what he really lacks is the confidence to make the decision.
The best leaders I worked around wanted information. They wanted people in the room who disagreed with them. Pete Carroll was constantly asking questions, searching for different approaches, and looking for anything that might improve the organization. But eventually the meeting ended. Practice had to be scheduled. The roster had to be set. A player had to play or sit. A game plan had to become the game plan. Someone had to make the call. AI can make that leader better informed before the decision. It cannot provide the courage required to make it or accept responsibility afterward.
The Leader Still Owns the Decision
We are going to hear extraordinary claims about what artificial intelligence will replace. Some of them will prove correct. Jobs will change. Certain tasks will disappear. Organizations will become more efficient. Leaders who refuse to adapt will struggle against competitors who learn to use these tools effectively. But leadership has never been defined by who could process information fastest.
Leadership is deciding what matters when everything appears important. It is understanding the people affected by a decision instead of only studying the numbers attached to it. It is recognizing when the available evidence is incomplete, asking the question nobody else has asked, and having enough experience to notice when an answer does not pass the smell test. Most importantly, leadership is ownership.
AI cannot stand in front of the employees whose jobs were affected by a decision. It cannot rebuild trust after a strategy fails. It cannot look an athlete in the eye and explain why he is not playing. It cannot accept responsibility in front of a board, a locker room, a customer, or an organization. The leader can. As AI gives more people access to increasingly similar information, that responsibility becomes more visible, not less. The advantage will no longer belong simply to the person who possesses the most information. It will belong to the person who knows what information to trust, what questions to ask, what information is missing, when enough information has been gathered, and what decision needs to be made.
AI may eventually make information cheap. Judgment will become more valuable because of it.
CoachC Insight
Every major step forward in technology changes the tools we use to do our jobs. AI will be no different. Leaders should embrace its ability to research faster, identify patterns, challenge previously held ideas, and show us possibilities that might have been missed. Refusing to use those capabilities because we are uncomfortable with change is no more intelligent than blindly accepting everything the technology produces. The challenge is understanding where the tool ends and leadership begins.
AI is not the problem. Bad leadership is. If leaders blindly accept what AI tells them simply because the answer comes quickly and sounds convincing, we are going to have problems. The best leaders will use AI differently. They will use it to challenge their thinking, sharpen their judgment, expose what they may have missed, and enhance the skills they have spent years developing. They will understand that AI is a powerful tool, but it is still a tool. The danger begins when leaders allow it to replace the knowledge, experience, judgment, and responsibility that leadership requires. AI will not make every leader great. It may simply make it much easier to see which leaders already understand how to lead.
Teachable Reminders
More information does not automatically create better judgment.
AI should challenge your thinking, not replace it.
The quality of an answer is often determined by the quality of the question.
Technology can identify patterns without understanding every circumstance surrounding them.
Strong leaders use information that disagrees with them instead of searching only for confirmation.
There comes a point when analysis must become action.
You can delegate research and analysis. You cannot delegate ownership of the final decision.
Application Questions
· Are you using AI to expand your thinking or allowing it to do your thinking for you?
· When AI gives you an answer, do you know enough about the subject to recognize when something does not make sense?
· How often do you ask AI to challenge your position rather than support it?
· Are there decisions in your organization where additional information has become an excuse for avoiding action?
· Does your team understand where AI's responsibility ends and human accountability begins?
· If a decision based partly on AI turns out to be wrong, who in your organization owns it?
