Q&A with Stanford professor Erik Brynjolfsson on generative AI, potential productivity gains, his “J-curve” and “Turing trap” concepts, AI drawbacks, and more
The Stanford professor on what generative AI will mean for productivity, jobs and the society of the future X: @tejparikh90 , @michaelrstrain , @futureworkinst , @pawlega , @tonytassell , @erikbryn , and @andrewtghill X: Tej Parikh / @tejparikh90 : My interview with @erikbryn for the @ft on what generative AI will mean for productivity, society and our institutions: ‘This could be the best decade in history — or the worst’ https://www.ft.com/... Michael R. Strain / @michaelrstrain : .@erikbryn is one of the most expert and thoughtful voices on the economics of technology. I always learn from him. Erik Brynjolfsson: ‘This could be the best decade in history — or the worst’ https://www.ft.com/... @futureworkinst : “I'm betting that productivity growth is maybe significantly higher than the Congressional Budget Office is projecting. They projected 1.4% average per year. I think it could be twice that.” @erikbryn in @FT talking productivity benefits of #AI. https://www.ft.com/... @pawlega : We've seen something like a J-curve with earlier general purpose technologies like the steam engine, electricity, and early computers. My reading of the evidence is that it will happen faster with AI https://www.ft.com/... Tony Tassell / @tonytassell : Views on AI are often divided between the doomers and the evangelists. Here Stanford Professor @erikbryn talks with @tejparikh90 on the huge potential and the risks. He says AI could double average US productivity growth. https://www.ft.com/... Erik Brynjolfsson / @erikbryn : Thanks, Michael. @tejparikh90 did a great job asking questions so we covered a lot of territory. The coming years will be very consequential for human flourishing. Andrew Hill / @andrewtghill : Excellent stimulating exchange on gen-AI's potential, between @erikbryn and @tejparikh90: ‘On the productivity side, I think 2024 is going to be a year of harvesting a lot of the capabilities’ https://www.ft.com/... via @ft
Context & Ripple Effects
Brynjolfsson’s interview connects the generative-AI debate to the economics of general-purpose technologies: capability gains may precede broad productivity gains while organizations redesign work and institutions adapt. It follows a period in which AI regulation and social risk were central policy concerns alongside industry discussion.
The significance is the attempt to link near-term deployment with longer-run outcomes for jobs and institutions, rather than treating model capability as a sufficient measure of economic impact. That framing also contrasts with a later view that AI alone may deliver only limited productivity gains unless it enables new activity.
First-order effects
- The interview gives executives and policymakers a framework—the J-curve—for interpreting a lag between generative-AI investment and measured productivity, while arguing that the lag could be shorter than for earlier technologies.
- Brynjolfsson’s higher productivity-growth scenario raises the stakes of 2024 deployment decisions, but his Turing-trap warning keeps job, institutional and societal risks within the same decision frame.
Second-order effects
- Companies seeking to “harvest” generative-AI capabilities will face pressure to change workflows and complementary practices, not simply add models; otherwise the expected productivity gains may not appear in aggregate data.
- The gap between optimistic productivity scenarios and regulatory concern increases pressure on policymakers to balance diffusion of AI tools with safeguards for their labor-market and social effects.
Third-order effects
- If the J-curve pattern holds, AI’s economic impact will be determined increasingly by organizational adoption and institutional response rather than headline model advances alone—an instance of AI moving into cognitive-task workflows.
- The Turing-trap framing suggests that the distribution of AI-driven productivity gains may become a central structural question: strong output growth would not by itself settle outcomes for jobs or society.
The trend: Generative AI is shifting from a model-capability story toward a contest over whether complementary organizational change and policy can convert adoption into broadly shared productivity growth.