No, the AI Is Not Neutral

"The AI is neutral — it's just math." I hear some version of this in almost every AI conversation, and it's the most expensive myth in AI adoption.

Math is neutral. But an AI system isn't just math. It's math plus the data someone chose, the labels someone assigned, the objective someone defined, and the cutoff someone set. Every one of those choices reflects human judgment — including human blind spots.

In plain English: the neutrality is in the arithmetic. Everything around the arithmetic is a decision someone made.

Why does the myth matter? Because "it's just math" ends conversations. It tells teams they don't need to test for fairness ("the numbers don't discriminate"), don't need oversight ("the system is objective"), and don't need to document choices ("there were no choices"). Each of those is wrong, and each gets more expensive the longer the system runs.

Replace the myth with a better sentence: "The AI reflects the choices behind it." Then go examine the choices — the data, the labels, the thresholds. That's where fairness lives, and it's entirely within your control.

I explain frameworks; I don't give legal counsel.