The Alignment Problem: Machine Learning and Human Values

Why this book is here

AI systems can optimize metrics that do not match what humans actually intend. An accessible exploration of the gap between machine-learning objectives and human values — grounding the governance and risk layer: why metrics, rewards and delegated authority cannot be blindly equated with human intent. The interactive demo stays educational and fully synthetic.

Use at work

Use at work: in the Living Organization, let the visitor see the consequence of choosing the wrong metric or removing the human confirmation — as a clearly fictional, educational scenario.