AI Boss

Welcome to the documentation portal for AI Boss: When Artificial Intelligence Manages People in European Enterprises and Public Authorities, Book 7 of The Architecture as Code Series by Gunnar Nordqvist. This site publishes the manuscript in a web-friendly format so that you can explore the chapters, diagrams and appendices directly in your browser. The source files are available in the repository at GitHub.

About the book

The earlier volumes of the series made architecture, governance, the AI platform and cyber defence executable, examined the evidence for AI as a threat, and described how the security architect's role is transformed when agents act. AI Boss asks what happens when artificial intelligence takes the manager's chair: when systems assign, pace and instruct work, measure and rate people, shape hiring, pay, discipline and dismissal, and when people in turn become the managers of AI agents. Algorithmic management is not a forecast. It is observable today on delivery platforms, in warehouses, in hiring and, increasingly, in the office.

The book divides management into five functions: M1 Direction, M2 Evaluation, M3 Consequence, M4 Development and M5 Care. Its thesis, which it tests against the evidence and judges in its conclusions, is that AI may direct work but may not, on its own, decide about people. The functions that assign and pace work can be delegated to a degree matched to reversibility; the functions that carry consequences for a person always have a named human decider who can be identified, contested and held to account, and Level 4 Autonomous is never used for M3 Consequence. Every management decision that AI makes or shapes about a person becomes a record the person can see, contest and have corrected: the Management Decision Record. The manager's role is transformed, not abolished: from supervisor of tasks to accountable leader of people and conductor of agents.

The book is written for line managers and the executives they report to, for HR directors and practitioners, for works councils, unions and other workers' representatives, for enterprise, solution and AI architects who build or procure the systems that manage people, for data protection officers, legal, risk and audit functions, and for public authorities as employers and as regulators. It covers the AI boss as it exists today, what it does to people, what changes when the team is software, Europe's legal answer, and a design for an accountable AI boss, followed through a year at two fictional organisations: Nordvik Region, a regional public authority whose rostering system plans the shifts of its hospital and home-care staff, and Hansa Mutual, an insurance group whose claims-triage agent already allocates work to people and whose team leaders are beginning to direct agents. The book is defensive and constructive: it describes monitoring at the level needed to assess and govern it, and never as a way to watch people covertly or to bypass their representatives.

The Architecture as Code Series

  1. Architecture as Code, published at aac.geon.se, supplies the underlying discipline.
  2. AI Governance, published at aig.geon.se, covers the regulatory map, risk management, governance as code, assurance, organisation and leadership.
  3. AI Platform and Security, published at aips.geon.se, covers the platform, DevSecMLOps, infrastructure, security and operations.
  4. AI Red Team, published at aired.geon.se, puts AI to work as an active part of cyber defence.
  5. AI Threat, published at aithreat.geon.se, examines the evidence for AI as a threat to people, society and control.
  6. The Transformed Security Architect, published at tsa.geon.se, describes what the security architect role has become and must become.
  7. AI Boss (this book) examines what happens when AI manages people, and how to keep a named human accountable for decisions about them.

Each book stands on its own. Where a topic belongs to an earlier volume, this book summarises it and refers to that volume by chapter number.

How to use this site

  • Start with the Introduction for the question, the thesis, the five management functions in brief and the two organisations that run through the book.
  • Follow the parts in order: the question, the AI boss today, what it does to people, managing machines, Europe's answer, designing the accountable AI boss, and experience and verdict.
  • Use the appendices for the glossary, the code examples, the templates and tools, and the bibliography.

The book has 33 chapters in seven parts. The Book Blueprint records the brief and status of every chapter, and Series Foundations records the conclusions from the earlier volumes that the book builds on. Contributors should also read the Chapter Authoring Guide.

Suggestions and corrections are welcome as issues or pull requests in the repository.