Predictable Volatility: Why Meaning — Not Models — Determines Whether AI Breaks You or Builds You

"Without meaning, there is no trust."— Valerie A. Logan, Founder, The Data Lodge Former VP Advisor to CDOs, Gartner"Anyone authorising AI systems at consequential scale should read the Override Problem chapter before signing off on another deployment."— Jon Hammant, AWS UK/I Specialist Leader & CTO "One of the most important leadership books I've read in years."— Dr Debbie Bayntun-Lees, Professor of Organisation & Leadership Development, Hult International Business School (Ashridge)Most organisations don't collapse into chaos. They collapse into coherence.Signals align. Dashboards look stable. Governance concludes correctly. And the failure that was already in motion continues — because no one was authorised to question what the system meant.Predictable Volatility examines why intelligent organisations keep producing the same failures, and why deploying more AI makes the problem worse rather than better. Drawing on fourteen forensic cases — including the UK Post Office Horizon scandal, Knight Capital, Boeing MCAS, Volkswagen, Wirecard, Robodebt, and the Dutch Toeslagenaffaire — Andrew J. Turner identifies a structural pattern that predates AI and is now being industrialised by it.The failure is not in the models. It is in the layer between data and decision, where meaning is assigned, where assumptions are frozen, and where the authority to stop execution quietly disappears.Turner introduces the Interpretation Gap: the distance between what a system is doing and what the organisation can understand, contest, and act upon in time. He develops the Meaning Mesh — the four conditions that keep interpretation honest — and the concept of Decision Architecture: the practical arrangement of roles, rights, and stopping power that determines whether human judgment can interrupt automated execution before consequences compound.This is not a book about technology risk. It is a book about the structural conditions that make failure foreseeable — and the design choices that make it preventable.The system this book describes is not coming. In most organisations, it is already running. Read more