According to a recent article in Harvard Business Review, AI is starting to undermine leaders’ judgment. The issue is not simply that people are using AI badly. It is that AI makes it dangerously easy to stop thinking independently.
Original thinking depends on two things: seeing beyond the obvious and forming your own interpretation of what you see. In practice, that means noticing weak signals, adjacent patterns, and uncomfortable anomalies, while resisting the temptation to outsource the conclusion. That is also one of the central arguments in my latest book, Stop predicting: 13 lessons from 1,400 attempts to see what was coming.
Sameness is the warning sign
Most innovation teams now have access to broadly the same AI models, similar prompt libraries, similar productivity tools, and similar access to information. Yet the results are wildly different. Some teams are using AI to widen their thinking, explore possibilities, and accelerate experimentation. Others are producing an avalanche of competent, polished, and utterly forgettable ideas.
That should tell us something important. The competitive advantage is not the model. AI amplifies the innovation system you already have. If your organisation is curious, diverse, challenging, and willing to question assumptions, AI can make that system more powerful. If your organisation is cautious, politically constrained, internally homogeneous, and addicted to safe answers, AI will scale that too. AI does not remove the human bottleneck. It exposes it.
Do not industrialise mediocrity
There is a bigger risk here than bad prompting. When people begin to defer to AI-generated explanations because they sound fluent, logical, and complete, they can gradually stop forming their own view. That is where the problem becomes structural.
If your people all start from similar assumptions, ask similar questions, and rely on similar models, convergence becomes inevitable. The answers become cleaner, faster, and more convincing, but also more average. This is how mediocrity gets industrialised: not because AI is unintelligent, but because human judgement becomes passive.
The machine becomes the first thinker rather than the second. That is the mistake.
The real advantage is judgment
In Stop Predicting, one of the recurring lessons is that the advantage does not come from predicting the future more accurately. It comes from becoming a better observer: scanning more widely, mixing unrelated signals, looking for what does not fit, challenging assumptions, asking better questions, and forming a point of view before the answer becomes obvious.
AI can support all of that, but only if the human being remains intellectually active. AI does not have to make us stupid, but it makes surrendering our judgment extraordinarily convenient. Once everyone works from the same models, the same assumptions, and the same plausible answers, mediocrity can scale at machine speed.
Use AI second
One of the simplest changes leadership teams can make is to alter the sequence. Do not immediately ask AI what it thinks. Think first. Form a hypothesis, take a position, and make your assumptions visible. Then ask AI to attack them. Ask it what you missed. Ask for the opposing view. Ask which weak signals contradict your conclusion. Ask it to stress-test the logic and show you where you may be wrong. Used this way, AI becomes a thinking partner rather than an answer machine. There is a huge difference.
Teach to disagree
Do not just teach people how to prompt AI. Teach them how to disagree with it. Widen the signals, form a view first, introduce deliberate dissent and encourage uncomfortable perspectives. Then use AI to challenge, extend and stress-test the thinking rather than to supply it. Because the real risk is not that AI becomes smarter. It is that we become more average.
