nber.org faviconErik Brynjolfsson, Daniel Rock, Chad Syverson·nber.org·

Artificial Intelligence and the Modern Productivity Paradox: A Clash of Expectations and Statistics

Key Takeaway

Addresses the puzzle of why rapid AI progress hasn't boosted productivity growth yet, proposing a J-curve of intangible asset development.


Brynjolfsson, Rock, and Syverson's exploration of the modern productivity paradox is an incredibly comforting read for anyone building in the trenches. They address the glaring contradiction between our daily experience of dizzying AI breakthroughs and the flatlining productivity statistics of the broader economy. Their explanation—the 'Productivity J-curve'—argues that general-purpose technologies require massive, unseen investments in intangible assets, business processes, and human capital before their economic benefits show up in the numbers. As an engineer, this makes total sense. We know that deploying a new model is just five percent of the battle; the other ninety-five percent is the hard, unglamorous work of refactoring legacy systems, clean-up pipelines, and user retraining.

As a founder, the J-curve concept is a crucial framework for managing expectations and capital. It means we are in the 'investment' phase of the curve, where companies are spending heavily to integrate AI, but the transformative organizational changes haven't fully matured. Startups that succeed in this environment won't just build the fastest models; they will build the software and workflows that help traditional enterprises navigate this adaptation gap. The productivity boom is coming, but it requires us to build the bridging infrastructure that turns raw capability into realized organizational value.

What stuck with me

  • The J-curve lag: Significant lag exists between the introduction of a general-purpose technology and its measurable impact on macroeconomic productivity statistics.
  • Intangible asset bottleneck: The real cost of adopting artificial intelligence lies not in purchasing software licenses, but in redesigning business workflows and training employees.
  • Bridge product opportunities: The most valuable startups in this epoch are those building the integration layer that simplifies the complex transition for traditional businesses.

Discussion & Comments

Have thoughts on this recommendation? Share your perspective below. Comments are reviewed before they appear.