Ben Moll and Alex Imas argue in J. Bradford DeLong’s crosspost that AI-led growth may be slower

Ben Moll and Alex Imas argue in J. Bradford DeLong’s crosspost that AI-led growth may be slower

Originally published on October 18, 2026, J. Bradford DeLong’s page crossposts an argument by Ben Moll and Alex Imas, then adds DeLong’s own response. Moll and Imas argue that AI could greatly increase economic output without producing double-digit annual growth in the next 10–15 years. Their central point is that automating tasks does not remove limits on what people want to buy, what firms can build, or how quickly technology spreads. Spending may shift toward services where people value human involvement, while physical work, energy, construction and political resistance can slow automation. They also argue that weak demand, cyber incidents and the difficulty of automating research could constrain growth. Their conclusion is that a large rise in GDP over a decade or two is more likely than double-digit growth soon; they say 4–5% growth would already be massive. DeLong calls even 5% productivity growth a huge shift, but says Moll and Imas are “way optimistic.” He also argues that discussion often conflates distinct questions, including whether AI changes human life, creates jobs and businesses, or brings financial returns. He describes current large language models as a linguistic veneer over retrieval, and says engineering success has outpaced understanding of how they work. How much AI raises measured GDP, and how its broader effects unfold, remain uncertain.

Publication: 9e90d1d2-5506-9693-502c-ad8db454e24f

attribution — pass: The reading correctly identifies the page as DeLong's crosspost of Ben Moll and Alex Imas's argument, followed by DeLong's own response. It keeps the arguments of the two sources separate.

Brad DeLong here: Nevertheless, 5% productivity growth is a huge shift.

summary — pass: The summary leads with the actors and the central thesis: AI may raise GDP levels a lot, but double-digit growth is unlikely within 10–15 years. It conveys the key constraints in plain language: demand shifting to human services, physical work, energy, politics, weak demand, cyber incidents and research automation. It keeps DeLong's dissent and his separate LLM skepticism. LLM is defined in the definitions, and the question is neutral and follows from the evidence.

Our claim is about the timeline: in the next 10-15 years, growth rates upwards of 10% are extremely unlikely and a much more reasonable baseline is, say, 4-5% (which would already be massive).

date — pass: The October 18, 2026 date matches the UTC day of evidence.originalPublished and the page dateline. The Sep 09, 2026 date belongs to the original Moll and Imas post, which the reading does not misstate. Forecasts are presented as the authors' arguments, and there is no hindsight framing. This is a publisher page, so no archive disclosure is needed, and modifiedAt equals the publication time.

Brad DeLong Oct 18, 2026

support — pass: Each claim matches a verbatim excerpt. This includes the 4–5% baseline, the constraint from non-automated demand, the authors' own caveat on weak empirical evidence for demand effects, DeLong's 'way optimistic' view and his claim that discussion conflates questions, and the 'linguistic veneer' characterization. No benefits, costs or forecasts are invented, and these are grounded as source assertions rather than verified facts.

If there is indeed demand for goods and services that are not automated, growth is constrained by this part of the economy, which is unaffected by the AI capabilities explosion.

media — pass: No media candidates were supplied and none was selected, so there is no identity or relevance problem.