The SCAR Framework
A Systematic Approach to AI Decision-Making in Critical Systems
Most AI failures are not technical. They are categorical: organizations put AI on problems it cannot solve and keep it away from the ones it can. This book gives you a repeatable way to tell the difference before you deploy, and to document the decision so it holds up when someone asks why.
Inside the book
-
Why AI projects fail, and the pattern behind it
From Watson for Oncology to the fatal Uber autonomous vehicle crash, the failures share a shape. The book traces that shape and gives you the assessment questions that expose it before money and credibility are committed.
-
Four dimensions that replace guesswork
Safety, Complexity, Accountability, and Resilience turn an AI decision from a bet into an evaluation. You get the complete worksheets, scoring matrices, and decision trees, and a clear reading of when AI belongs in your context and when it does not.
-
What aviation teaches about automation
Air France 447, the Boeing 737 MAX, Tesla Autopilot, and Garmin Autoland show why sophisticated systems fail while deliberately simple ones save lives, and why certification authorities still keep machine learning out of flight control. The lessons transfer to any industry where failure means litigation, bankruptcy, or death.
-
Documentation that protects the decision, and the decision-maker
When an AI system fails, investigators, regulators, and lawyers will ask what you knew and when. The book lays out what to record, the red flags that should stop a deployment, and how to tell leadership that AI will not work when that is the honest answer.