The biggest problem in company AI roll-outs right now isn't hallucinations.

It's resulting.

Annie Duke (champion poker player turned author) has a name for the mistake most leaders are making with AI.

Resulting: judging a decision by its outcome instead of the process that made it.

So far I've watched resulting in AI play out in two ways:

One: An AI chatbot disappoints a customer. See I knew we were wrong to embrace AI!

Two: A promising demo app becomes a new religion. OMG stop the presses: I’m replacing every employee with AI right now!

Neither reflect the right way to think about the situation.

Duke's point: Life isn’t chess, it’s poker. In chess, there is a right answer. In poker and business, hidden information and luck mean a brilliant decision can blow up, or a terrible one can pay off.

In Never Split the Difference, former FBI hostage negotiator Chris Voss calls hidden information Black Swans: pieces of information that, once uncovered, completely reframe everything. Every AI deployment is full of them: edge cases the demo never hit, user behaviors no one modeled, and exciting possibilities that don’t reveal themselves for months.

Your AI strategy is one Black Swan away from being either a crisis or a breakthrough. Certainty, in today’s environment, isn't strength. It's legerdemain: sleight of hand that fools you as much as your audience.

Duke's suggestion is simple: separate the quality of the decision from the quality of the outcome. Before you do/don’t do anything, write the bet. What do you believe will happen? How confident are you, in an actual percentage? What would change your mind? Then run the pre-mortem: assume it fails. Why?

Charlie Munger would call this inversion. Voss would call it hunting for Black Swans. Duke would call it calibration.

Most companies seem to be skipping this process entirely. They're asking "did it work?" when they should be asking "were we right to believe it would?"

Nobody knows which AI bets will pay off. Anyone who says otherwise is selling something.

Are you aware of the bets in AI you’re making right now? And are you aware that not embracing this technology is, itself, a bet?