The AI Dichotomy

Capability and perception get the attention. Consequence decides who pays.

Every conversation about AI fixates on two questions: what can it do, and what do people believe it can do. The AI Dichotomy adds the one that actually decides outcomes - what happens, and to whom, when the system is wrong.

Air Canada learned it when a tribunal held the company - not its chatbot - liable for a wrong answer. Zillow learned it when a confident algorithm met a messy market - a write-down of more than half a billion dollars and roughly two thousand jobs cut. IBM spent years marketing an AI that could fight cancer while its own documents told a different story. The pattern repeats across every failed deployment: AI value collapses not simply when capability and perception drift apart, but when no one has asked who carries the weight of a wrong answer.

Through a single team building, deploying, and finally over-trusting an AI system, you'll learn to read any AI claim and find the source of the smoke; tell a demo that impresses from a deployment that delivers; see why pilots succeed and rollouts quietly fail; decide when to let an AI system act, and how to match its autonomy to the consequence of being wrong; and govern AI by naming who owns the outcome - before the failure, not after.

It is not a book about the future of AI. It is a field guide to the present, written for the people who actually have to make AI deliver.

If you have ever watched an impressive demo and quietly wondered what happens when it's wrong, this book is for you.