Cost Of Unwanted Intelligence

Every board on earth asked the same question in the same quarter: what is our AI strategy? Three years later the answers are arriving, and they look like invoices nobody modelled. Uber burned an annual AI budget before mid-year. Microsoft pulled internal licences after discovering engineers were running up two thousand dollars a month each — in usage, not salary. Amazon built a leaderboard ranking staff by AI consumption and watched employees generate expensive nonsense to climb it. And MIT found that roughly ninety-five percent of enterprise AI spending had produced nothing measurable at all.

Cost of Unwanted Intelligence explains why. Its argument is a single sentence — the price of a rented mind is set not by the meter, but by the gap between the intelligence you hire and the work you actually have — and it earns that sentence story by story.

Written for leaders who read a P&L rather than a GPU bill, the book reframes AI as labor: a tireless genius with no memory, who occasionally lies with total confidence, and whose paycheck grows every time it opens its mouth. Six chapters name the six costs that sit beneath the API invoice — Delegation, Submersion, Appetite, Wrongness, Last Mile, and Decay. The final act assembles them into the Intelligence Ledger, calculates the Unwanted Intelligence Tax, and gives the reader three levers: Deploy, Narrow, Retire.

The contrarian punchline: the cheapest intelligence is the kind you stopped renting.