A portfolio CFO walks into Monday's review with an AI-built variance model the team never asked for.

The room nods along for forty minutes. Not one person tests a single assumption, because the deck already sounds like someone tested it for them.

A creative director, reported in Chief Executive, picked up an odd habit at a former job leading creative teams at Disney: a head tilt whenever an idea landed well in a meeting. They never said a word about it.

Their team mapped the tilt within weeks and began pitching toward it instead of the problem on the table. A colleague finally pointed out what the director's own meeting notes had become: a record of head tilts, not arguments.

The director stopped taking notes. The pitches got stranger and sharper within days, because the room had stopped reading a face and started reading the work again.

Wharton ran a controlled version of this: give people either a correct AI answer or a wrong one delivered with total certainty. Wrong answers still dropped accuracy 15 points below working alone, and people's certainty rose anyway.

A separate, smaller MIT Media Lab study of 54 writers found the same direction: the more a person deferred to the tool, the weaker their own engagement with the work.

A twelve-person services firm runs on the same physics. The founder who always sounds sure trains three managers to stop raising the objection that would have saved the client relationship, and the firm only finds out once the invoice gets disputed.

Timing matters more than how sure you sound.

State the part you are unsure about before the room reacts to the polished version in front of it, and name one specific person to argue the opposite case before the meeting ends (the newest hire on the team is usually safest for this).

What happens to the pitch that never gets challenged? It gets funded anyway, and the bill for sounding sure arrives at the next review, not this one.

But the fix starts with you, not them.