\cleardoublepage \part{The AI Boss Today} \setbookpart{The AI Boss Today}

Part II: The AI Boss Today

Part I defined the AI boss in the abstract. This part shows where it already works, sector by sector, and what it does in each place. The evidence is uneven: platform work and warehouses are well documented by researchers, courts and regulators, hiring systems by audits and litigation, and the office by vendor reports and early studies. Each chapter therefore separates what is observable and documented from what is reported by the vendors that sell the systems, and states what is still unknown.

The common pattern across the six chapters is that delegation rarely happens by an explicit decision. A scheduling module is switched on, a performance dashboard is shared with team leaders, a screening service is added to the applicant tracking system, an office assistant begins to summarise who did what. Each step looks like a tool for the manager. Taken together they move parts of M1 Direction and M2 Evaluation to the system, and they shape M3 Consequence through the scores and flags on which managers act. Few organisations can say, for a given system, which functions it performs and at which degree of delegation. Part II builds the evidence for why they should.

Chapters in this part:

  • Chapter 5, Platform Work: The Algorithm as Dispatcher, examines dispatch, pricing, ratings and deactivation in ride-hailing, delivery and online labour platforms, where the AI boss is most complete and where courts and the Platform Work Directive have intervened.
  • Chapter 6, The Warehouse and the Shop Floor: Pace, Rate and Monitoring, examines rate targets, time-off-task measurement and automated warnings in logistics and manufacturing, and the regulatory and inspection responses to them.
  • Chapter 7, Hiring by Machine: Screening, Scoring and Automated Interviews, examines CV screening, assessment games and recorded interviews, the evidence on bias, and the line between ranking applicants and deciding who is rejected.
  • Chapter 8, Performance Scores, Productivity Metrics and Dismissal by Algorithm, examines how scores become ratings, warnings and terminations, and where the human decider disappears from the chain.
  • Chapter 9, Scheduling, Pay and the Allocation of Work, examines rostering, dynamic pay and the allocation of tasks and hours, including in health and social care, where Nordvik Region's rostering system, ShiftWise, sits.
  • Chapter 10, Knowledge Work: The AI Manager Enters the Office, examines office assistants and agents that assign tasks, summarise contributions and report on colleagues, and the claims-triage agent that allocates claims to claims handlers at Hansa Mutual.

Management functions in this part. The chapters concern mainly M1 Direction and M2 Evaluation, which are where delegation to systems has gone furthest, and M3 Consequence, which is where the harm and the litigation concentrate. M5 Care enters wherever pace and monitoring affect health and safety (Chapters 6 and 9), and M4 Development wherever feedback is reduced to a score (Chapters 8 and 10). Every chapter names the functions it concerns in its opening and its Summary, and closes with a table across the usage modes or the degrees of delegation.

The thread from Part I. Part I gave the reader five functions, five degrees of delegation and six usage modes. Part II applies them to real systems, so that each sector can be described by the same profile: which functions the system performs, at which level, in which mode, and who answers for the consequence. The result is a map of the AI boss as it exists, not as it is promised or feared.

What you will learn in this part:

  • Where algorithmic management is already in use in Europe, and how well each use is documented
  • How to describe any system that manages people by its management functions and degrees of delegation
  • Where courts, regulators and inspectorates have already drawn lines
  • How Nordvik Region and Hansa Mutual meet each pattern in their own operations

Part III turns from what the AI boss does to what it does to the people it manages.