Capability and Adoption

This paper opens the AI in the Enterprise series by framing what artificial intelligence actually does to firms, economically and operationally, before the later papers turn to safety, governance and implementation. Brynjolfsson and McAfee treat AI as a general-purpose technology reshaping the economy — as consequential as steam or electricity, and, like them, slow to show up in the productivity figures until organisations make the complementary changes around it. Agrawal, Gans and Goldfarb reduce the technology to a single economic effect: a collapse in the cost of prediction. Iansiti and Lakhani show how that collapse rewires the firm's operating model and competitive dynamics through the idea of the AI factory.

The three lenses are complementary rather than competing: general-purpose technology explains why adoption is slow and organisation-wide; cheaper prediction explains where the value actually shows up; the AI factory explains what has to be rebuilt — data, algorithms, experimentation and infrastructure operating as a single system — to capture it.

For the operator, the through-line is decisive: the value is not the model but the redesign of work and systems around cheap prediction and augmentation, and advantage accrues to the complements — judgement, data and process — rather than to the technology itself. The next paper in the series turns to the risks that arrive alongside the capability.