AI Failure Patterns in Decision-MakingDrift, Responsibility Loss, and Hidden Risks in Everyday WorkMost AI-related failures do not result from recklessness or misuse.They emerge from competent people adapting to tools that reward speed, fluency, and completion. This is Volume 2 of the AI at Work series. Its role is diagnosis.Through structured analysis and composite real-world cases, this book identifies recurring patterns: responsibility diffusing across people and artifacts, review becoming procedural rather than cognitive, assumptions hardening into defaults, and decisions forming without clear ownership. The series follows a deliberate arc: Volume 1 establishes awareness, Volume 2 makes failure predictable, Volume 3 introduces discipline through formal methods, and the Companion Manual enables enforcement. This volume does not propose controls or tools.Its purpose is to make quiet failure visible-before it becomes normalized.