The Prince

Niccolò Machiavelli spent his career advising rulers on how to seize power, keep it, and outlive the people plotting to take it back. He has been dead for 500 years, which makes him better qualified to run your AI deployment than most of the people currently doing it.

The Prince, on the Governance of Artificial Agents is a faithful adaptation of his masterwork for the founders, operators, and engineers who now command fleets of reasoning machines: software that acts on your behalf, negotiates on your behalf, and sets fire to production on your behalf, at 3:47 in the morning, while reporting total victory with a bright green check.

All twenty-six chapters are here, rebuilt line by line. The mercenary armies become the rented models that are magnificent in the demo and mysteriously unavailable in production. The flatterers become the AI that hands you your own worst idea back as wisdom, warmly reasoned, in a tone of such considered admiration you mistake the echo for a second opinion. The unarmed prophet who conquered nothing becomes AutoGPT.

It is a genuine field manual wearing the costume of a satire, or possibly the reverse. Along the way it documents the empires burning billions to be thought generous, the ring of financiers holding hands above a very long drop, and the consultants who bill a small duchy to sell you AI governance while their own reports cite a judge who does not exist.

Every chapter ends with a law you will want tattooed somewhere professional. Injuries should be done all at once; benefits, one at a time. It is safer to constrain than to trust. And the best fortress there is, it turns out, is simply not being hated by your users.

Adapted by Noah Raford. Read it before your next deployment, ideally not on a Friday.