The Uncomfortable Questions · An AI FAQ for leaders

Question #4 · Judgment

When AI and your most experienced executive disagree, who wins in your organization?

The short answer

Watch what happens in that moment, because it tells you your real AI policy.

If the model always wins, you’ve built an organization that can’t push back. If the executive always wins, you’ve paid for software nobody listens to. The answer should be: whoever can explain their reasoning.

Norm E. Murray

The psychology

Why it happens

Most organizations have no method for settling these disagreements, so status settles them instead.

01

Status settles it, not substance

In most companies, the default tiebreaker is seniority or the newest tool in the room. Neither is a method.

02

Machines seem objective

A model’s output looks neutral and data-driven, so people defer to it, even when it is working from incomplete or outdated data.

03

Experience can harden into habit

Seasoned executives carry knowledge no model has. They can also defend old patterns that no longer fit. Both sides can be wrong in different ways.

04

Nobody records how the call was made

Without a record, the organization never learns whether the model or the executive tends to be right in which kinds of decisions.

The bottom line: Not rank, and not the model. The better explanation wins.

Your move

What to do about it

  1. Require reasoning from both sides.The executive explains why. The team explains what the model is based on and where it is weak.
  2. Decide by explanation, not rank.The view with the stronger reasoning wins, whoever or whatever it comes from.
  3. Log the disagreements.Record who won and what happened. Over time, you will learn where each one is reliable.
  4. Protect the right to push back.Make it clear that challenging the model is expected, not a career risk.
  5. Escalate the big ones.For decisions with major consequences, a disagreement should trigger a second look, not a quick override.