Human + Machine

Human + Machine grew out of the Accenture Research program Paul Daugherty runs, and it carries the strengths and limitations of that origin: it is data-rich, structured, and grounded in documented deployments rather than speculation, and it reads a little corporate in places where the underlying realities are messier than the framework admits.

The book's central thesis is that AI does not replace jobs so much as it shifts the responsibilities inside them, and the competitive advantage goes to the companies that redesign roles around what they call the "missing middle" — the space where machines handle routine execution and humans handle judgment, training, and accountability. Daugherty and Wilson introduce eight "fusion skills" that operate in that middle: training algorithms, explaining outputs, sustaining algorithms against adversarial input, and so on.

What makes it worth reading now is its taxonomy of how AI projects actually fail inside enterprises. The closing chapters cover the leadership moves that distinguish companies that achieve measurable ROI from AI from those that spend years on pilots that never reach production. For founders and operators at smaller companies, the book's framework translates cleanly: most small-team arguments about AI are really about which "fusion skill" the human should be owning, and which one the machine should be doing.