The Confidence Tax

I have sat in meetings about AI-native delivery with a calm face and a private feeling of, does anyone in this room actually know how this is supposed to work yet. I have also stood at the front of that same meeting and performed a certainty I did not have, because people were looking to me to have a handle on delivery and the calm face felt like the thing the job required. Both moves felt safe at the time. Both felt like competence. Only later did I notice that one of them had quietly cost me something, and that I had been paying it for a while without seeing the charge.

Performing certainty feels like the responsible choice, especially if you are the person others look to on delivery. Admitting you are partly improvising feels like losing the room, like confirming the quiet fear everyone is already carrying, that nobody is really driving this. So you reach for the confident take and the steady voice, and you move the meeting along, and everyone relaxes because you seem to know. The performance works in the moment. It does exactly what you wanted, which is to make the uncertainty disappear from view.

The pull to do this is strongest precisely when you are least sure, because that is when the gap between how much you know and how much you are expected to know is widest, and the performance is what covers the gap. Which means you reach for certainty hardest at the exact moment faking it does the most damage.

The performance stops you doing the one thing that actually helps, which is checking your assumptions instead of defending them. Once you have said the confident thing out loud, in front of people, you are committed to it. Now every new piece of evidence gets read through the lens of a position you have to protect, something to argue against rather than something to learn from. You have spent your energy defending a take instead of testing it, and defending and testing pull in opposite directions.

And the instinct you armoured with all that confidence gets more runway to do harm. A wrong call delivered with certainty is believed longer, acted on harder, and caught later than the same wrong call delivered with an honest question mark. The confidence does not make the call more right. It just removes the friction that would have surfaced the problem early, while the problem is still cheap to fix. That is the tax, and you pay it in the assumptions you never went back to check because checking would have meant admitting you were not sure.

"I am not sure yet, let us find out." The first time I said that to a team instead of performing a take, the relief in the room was immediate and obvious. Everyone had been holding the same uncertainty and privately assuming they were the only one not keeping up. One honest sentence did more for the work than a month of the confident version, because it put the real question on the table while there was still time to answer it, instead of burying it under a performance that everyone half-suspected anyway.

The people I have watched handle this year best are not the ones with the most assured take. They are the ones comfortable treating their own experience as a strong hypothesis rather than a settled fact, willing to be visibly unsure in front of a room. They are right more often, not despite the uncertainty but because of it, because being openly unsure is what keeps you checking. Certainty performed is contagious, and so is honesty, and only one of the two helps you actually work out what to do.

So, the last time you sounded certain in a delivery meeting: were you, or were you paying the tax?