Q&A with Anthropic co-founder Jack Clark on the parts of the economy AGI will affect last, why he's relatively bearish on AI-fueled economic growth, and more

And the inevitable future of AI-enhanced parenting — SUBSCRIBE - Apple - Google - Stitcher - Spotify - YouTube

Conversations with Tyler 2025-05-11 Tyler Cowen

Context & Ripple Effects

Clark’s current skepticism about AI-led growth extends a long-running line of inquiry: he previously discussed automation’s effect on white-collar work and AI research’s direction. This interview shifts that conversation from whether AI can automate tasks to which economic domains may remain resistant even under AGI.

The related coverage also shows the productivity question remaining unresolved, including a later debate over whether AI tools improve productivity and displace jobs. Clark’s emphasis on uneven effects and AI-enhanced parenting broadens the issue beyond workplace software adoption.

First-order effects

Clark’s relatively bearish growth outlook pushes back on treating AGI capability gains as an automatic proxy for economy-wide output gains; businesses and investors must distinguish technical progress from realized deployment and demand.
The focus on sectors AGI may affect last highlights that AI exposure will vary by activity, while AI-enhanced parenting brings consumer and family use into the economic and governance discussion.

Second-order effects

AI vendors and enterprise buyers face greater pressure to demonstrate measurable workflow gains rather than relying on broad AGI narratives, especially where adoption depends on trust, coordination, or physical-world execution.
As AI moves into parenting and other intimate consumer contexts, product design and policy debates are likely to center more on safeguards and accountability than on capability alone.

Third-order effects

If capability improvements continue to diffuse unevenly, AI’s economic impact may be shaped less by a single AGI threshold than by complementary institutions, distribution, and adoption constraints.
The larger structural question becomes whether AI creates broad productivity growth or primarily lowers the cost of selected cognitive tasks—a distinction still contested in the related coverage.

The trend: This is one data point in the shift from headline AGI forecasts toward scrutiny of the uneven, institution-dependent path from AI capability to economic value.

Discussion

@mattbeane Matt Beane on x
This is more or less where I am. Still, still, I wish for a workshop: get the 20-30%-ers in a room with the 3-5%-ers and hash it out. With data, prior research. Prereading required, serious business, no grandstanding. Building grounded answers, or at least plans for them.
@cowenconvos @cowenconvos on x
“I'm a technological pessimist who became an optimist through repeated beatings over the head of scale,” says Anthropic co-founder @jackclarkSF. And yet when estimating AI's impact on US economic growth, he suggests it's more likely to boost annual growth rates to 3-5% rather [vi…
@haydnbelfield Haydn Belfield on x
3-5% seems right to me - rich economies are really big! And would be remarkable, hugely positive growth rates for developed countries As I always caution though, you can get strategically destabilising technology from an additional few % growth
@jenmartinez Jennifer Martinez on x
This interview with @jackclarkSF and @tylercowen has been a long time coming - it was recorded at @AnthropicAI HQ back in March. Big thank you @tylercowen @Jeff_Holmes and @cowenconvos team for spending the afternoon with us!
@chrisgpotts Christopher Potts on x
This is a great interview! How, though, did Tyler miss the chance to find out whether the Anthropic logo is actually from, err, Breakfast of Champions? I thought he was going to ask around 55:18.

Chronicles