A wave of corporate experiments with generative AI has yet to yield the profits many investors anticipated. Despite heavy spending, an MIT Media Lab review finds that only a small share of pilots move beyond trials to deliver measurable gains, echoing Robert Solow’s “productivity paradox.” Executive confidence in AI rollouts is slipping amid scalability and integration challenges, even as some targeted back-office and marketing uses show promise. The historical record suggests a familiar pattern: general-purpose technologies often deliver gains only after complementary infrastructure, skills, and workflows evolve—a “productivity J-curve” with early friction and delayed payoffs. That timeline complicates today’s AI trade, where stretched valuations and bubble talk contrast with uneven operational results. For boards and investors, the message is to narrow use cases, invest in capabilities, and temper near-term expectations while positioning for longer-run productivity dividends.
Related articles:
Artificial Intelligence and the Modern Productivity Paradox: A Clash of Expectations and Statistics
The Productivity J-Curve: How Intangibles Complement General Purpose Technologies
Generative AI at Work
The Computer and the Dynamo: The Modern Productivity Paradox in a Not-Too-Distant Mirror
Productivity paradox
