AI is just another productivity tool and the productivity gains will be limited; for lasting economic expansion, AI must catalyze new industries and initiatives
Economic miracles stem from discovery, not repeating tasks at greater speed — The writer is author of 'How Progress Ends … Bluesky: @eicathomefinn and @dncampbell X: @mrrbourne , @kamilkazani , @carlbfrey , and @erikbryn Bluesky: Margot Finn / @eicathomefinn : ‘The shotgun marriage of the computer and the internet promised more than enhanced office efficiency—it envisioned a golden age of discovery....Yet research productivity has sagged.’ An astute analysis. Here's a sharing link for 3 non-subscribers: on.ft.com/43Karij Donald Campbell / @dncampbell : “Large language models gravitate towards the statistical consensus. A model trained before Galileo would have parroted a geocentric universe; fed 19th-century texts it would have proved human flight impossible before the Wright brothers succeeded.” www.ft.com/content/55bc... X: Ryan Bourne / @mrrbourne : These are complementary though, no? Freeing up people and resources for new innovative industries is one consequence of improving productivity in existing sectors. Kamil Galeev / @kamilkazani : I think our view on the relative importance of technologies is distorted, as we criminally underestimate significance of general purpose ones E.g. there has been no greater technological breakthrough than Haber-Bosch, at least in the last few centuries, yet it occupies exactly Carl Benedikt Frey / @carlbfrey : In the @FT today, I argue that if AI is just another productivity tool, the productivity gains will be limited. For lasting economic expansion, AI must catalyze new industries and initiatives. “Had the 19th century focused solely on better looms and ploughs, we would enjoy [image] Erik Brynjolfsson / @erikbryn : Carl is right. AI is already boosting productivity in many tasks, but the real power of general-purpose technologies, from the steam engine and electricity to AI, lies in catalyzing new products and processes.
Context & Ripple Effects
The argument sharpens a debate already framed by Erik Brynjolfsson’s “J-curve” view of generative AI adoption: task-level gains may arrive before organizations redesign products and processes around the technology. Here, Carl Benedikt Frey and the cited commentators set a higher bar—whether AI produces discovery and new economic activity rather than faster execution of existing work.
That distinction matters because the discussion is not primarily about model capability. It is about the conversion of a general-purpose technology into new products, processes, and industries—and the risk that automation alone leaves the broader growth effect modest.
First-order effects
- The piece shifts the near-term measure of AI’s value from isolated efficiency gains toward evidence of new initiatives, products, and scientific or industrial discovery.
- For companies deploying AI, the implication is that workflow automation can reduce effort but does not, by itself, establish a durable growth case; realizing one requires complementary process and business-model change.
Second-order effects
- AI vendors and enterprise buyers face pressure to demonstrate outcomes beyond generic productivity claims, favoring deployments tied to differentiated products or redesigned operations.
- The comparison raises the importance of organizational complements—expertise, experimentation, and implementation capacity—that determine whether task-level assistance becomes a new source of economic activity.
Third-order effects
- If adoption remains concentrated in automating established tasks, AI could resemble a broadly useful infrastructure technology whose benefits diffuse without creating a proportional wave of new industries or exceptional growth.
- If firms instead use AI to expand the feasible set of research, production, and services, competitive advantage may increasingly accrue to organizations able to turn models into novel offerings rather than merely deploy them cheaply.
The trend: AI’s economic significance is shifting from a debate over automation efficiency to a test of whether adoption catalyzes new industries and discovery-driven growth.