When Output Gets Cheap, Judgment Gets Expensive

Anyone can produce a fluent draft now. Very few people can tell if it's actually good.

‍

That's the shift AI has quietly forced. A competent memo, a plausible analysis, a polished recommendation, all of that can appear in seconds. So the scarce skill isn't generating the answer anymore. It's knowing whether the answer is true, useful, and right for the situation.

‍

Here's the catch: AI's help is uneven, and you often can't tell from the inside. In one field study of 758 consultants, those using AI were faster and sharper on tasks squarely inside the tool's range, but on a task just outside that range, they were more likely to land on the wrong answer than colleagues working without AI. The weak output looked exactly as confident as the strong one.

‍

Good judgment is what catches that gap. It's not a vague executive instinct it's specific and learnable: deciding which evidence is strong enough, which risk is real, which answer is merely the convenient one. It asks the annoying second question at the exact moment everyone else is speeding up.

‍

This article  is an extract of "The Human Skills Premium: Where Value Moves in the Age of AI" by Simon Blanchette, McGill University Desautels.