Hi {{first_name}}

I use AI most days.

It helps me test ideas, find gaps in an argument, and see possibilities I might otherwise miss. If I am working through a difficult decision, I can ask it to challenge my assumptions, construct the strongest case against my preferred option, and identify risks I have overlooked.

Within minutes, I can receive something that looks like a thoughtful second opinion.

Recently, however, I have wondered whether that experience contains a problem of its own.

AI can challenge my thinking without making me feel challenged.

That distinction matters more than it first appears.

When we ask another person to test our judgement, we give up some control over what happens next.

They may question the way we have framed the problem. They may notice that we are giving one risk more weight than another. They may hear certainty in our words but hesitation in our voice. They may know something about the people, history, or politics surrounding the decision that we have not considered important enough to mention.

They can also disagree in ways that are inconvenient.

AI is different. However sophisticated the response, the conversation begins inside a frame that we created. We choose the question, provide the context, and decide which facts belong. We can ask for objections, but we still determine which objections to pursue. We can end the conversation the moment we feel satisfied.

This does not make the analysis worthless. It makes it dependent on what we supplied and what we were willing to examine.

The danger is that the quality of the response can disguise the limits of the encounter. A decision may have been extensively analysed without ever being independently challenged.

The Center for Creative Leadership recently reported that 64 per cent of leaders say AI enables them to make more decisions without consulting other people. Faster access to analysis removes friction. However, contesting assumptions, adding context, and developing collective judgment occur within that friction.

Speed and convenience plainly account for some of AI's appeal. But there may be another attraction that is easier to miss.

Consulting another person involves exposure. They may see something we missed. We may have to defend a preference we cannot fully explain. Their disagreement may create tension, delay the decision, or leave us questioning our own judgement.

An AI system can offer the experience of consultation without the same interpersonal risk.

This does not mean that leaders are consciously using AI to avoid challenge. More often, the process simply feels efficient. The issue is what efficiency quietly removes.

The purpose of involving other people is not merely to obtain more answers. It is to encounter thinking that we do not control.

Not every decision needs a committee. Routine decisions should not require unnecessary meetings simply to prove that human beings were involved. AI can make individual judgement faster, broader, and better informed.

But consequential decisions deserve a different question:

Am I using AI to prepare for a better human conversation, or to make that conversation feel unnecessary?

Before deciding alone, it may help to consider who could challenge not only your proposed answer, but the question you have asked. Who understands the consequences you cannot fully describe in a prompt? Who is sufficiently independent to tell you that the problem may not be where you think it is?

The test is not whether AI has produced a convincing analysis. It is whether the decision requires judgement from someone who does not share your assumptions and cannot be directed by your prompt.

Answers are becoming abundant. Independent judgement is not.

AI can help us think more extensively. It cannot guarantee that we have stopped thinking alone.

Until next week,

Gavin

Further reading: Center for Creative Leadership, How AI Is Changing Decision-Making (2026)