
Prof @oiioxford Director, Future of Work @oxmartinschool at Oxford University. Author of How Progress Ends (2025) & The Technology Trap (@PrincetonUPress, 2019)
Joined August 2011
- Tweets3.3K
- Following545
- Followers7.6K
- Likes5.3K
Pinned Tweet
My latest in the @FT today. I make four points:
1) Capitalism's great achievement was moving activity out of the household and into the market — turning domestic production into paid specialisation, creating jobs, and making output visible to the national accounts. AI-enabled self-service might quietly reverse that centuries-long trend.
2) When a technology automates tasks within a service, it can trigger a Jevons paradox: the service gets cheaper, demand expands, and employment grows. When it lets people do the work themselves, demand for the service collapses.
3) AI extends this even to the manual trades, the supposed safe haven of the AI age. If a homeowner can ask a chatbot why their boiler keeps losing pressure, heating engineers lose call-outs.
4) When work shifts to the consumer, it vanishes from the economy statisticians measure. Replace a billing department with a chatbot and a firm records lower costs and higher output per worker; the national accounts register a productivity gain. But the hours patients spend decoding their own test results appear nowhere — not in labour statistics, not in GDP. As self-service spreads into professional domains, that blind spot will grow.
Carl Benedikt Frey retweeted
Are we in an intelligence explosion? This is one of the most important questions facing society right now. @lugaricano and I tried to put structure on this question and provide estimates based on currently available data.
METR recently reported that frontier capability growth roughly doubled after October 2025---but this is accompanied by annualized R&D spending growth of ~6x and technical staff ~3x over the same window. That highlights a crucial distinction between scaling laws--exponential capability gains bought with exponentially rising inputs--and an intelligence explosion (IE), where AI begins substituting for external inputs so each advance makes the next one easier.
We estimate several versions of a research production function (building on @ChadJonesEcon, @bfjo, and @Ph_Aghion) where a key parameter crossing a threshold signals an IE, while values below it correspond to an "ideas are harder to find," diminishing-returns world. Across specifications, our estimates sit below that threshold. Allowing for a break at METR's October 2025 cutoff points to a one-time step up in research effectiveness, with the rate of diminishing returns being nearly identical on both sides of the cutoff.
A few things to note. First, this exercise is not a forecast. It's an attempt to put structure on some of the important questions being debated right now: RSI, IE, scaling laws all depend on how outputs scale with inputs, and how feedback loops accelerate/decelerate this process. These estimates can change, and change very quickly.
Second, we are very explicit that the results are based on publicly available data; it is very much possible that METR had access to data that would change the estimates.
Third, we might be wrong! Let us know what you think:
aleximas.substack.com/p/are-…
Carl Benedikt Frey retweeted
AI means very different things depending on what a society already has. In rich economies, it competes with expensive human expertise; but when introduced elsewhere, it supplies expertise that was never available, @carlbfrey explains. project-syndicate.org/commen…
Carl Benedikt Frey retweeted
Post-Doctoral Researcher Vacancy: Join us at the University of Oxford! Our Economics, Inequality, and Opportunity team at INET Oxford is hiring a post-doc to start in September 2027 (or sooner) on a two-year contract. Deadline: December 1
Carl Benedikt Frey retweeted
Every time I think European voters cannot be more in denial about pensions, they prove me wrong: they are even more in denial than in my worst nightmare.
The average French person believes there are 6.4 workers for every retiree. The real number is about 1.8.
Carl Benedikt Frey retweeted
Who pays for AI? In a new working paper, we study this using banking data covering billions of transactions to see who pays for tools like ChatGPT, Claude, Gemini, etc. We find that 1-in-20 UK adults now pay for AI, representing a $1.4 billion market annually. (1/N)
Carl Benedikt Frey retweeted
.@carlbfrey calls for legislation to establish learning rights, akin to mineral rights or broadcasting rights. Each was created by statute when a newly valuable resource needed an owner before it could have a market. bit.ly/4xuVGLX
Carl Benedikt Frey retweeted
4/9 In the IT era, US multinationals in Europe got more out of IT than local firms under the same rules, mainly through people management (Bloom-Sadun-Van Reenen 2012). I suspect these management differences are still there, and matter for AI.
Carl Benedikt Frey retweeted
Quick thread on AI and productivity:
Lots of people say this chart shows AI is now showing up in the productivity data.
But I made this chart to argue the opposite: a trend that pre-dates AI and only shows up in the US doesn’t look like AI.
Here’s how I’m thinking about the data and underlying dynamics:
My latest in @ProSyn
AI means very different things depending on what a society already has. In rich economies, it competes with expensive human expertise; but when introduced elsewhere, it supplies expertise that was never available, @carlbfrey explains. project-syndicate.org/commen…
Carl Benedikt Frey retweeted
AI means very different things depending on what a society already has. In rich economies, it competes with expensive human expertise; but when introduced elsewhere, it supplies expertise that was never available, @carlbfrey explains. project-syndicate.org/commen…
Carl Benedikt Frey retweeted
The United States is both the home of the AI revolution and among the places where people like the technology the least. @carlbfrey considers why. project-syndicate.org/commen…
Carl Benedikt Frey retweeted
Since Zucman is saying "but we were peer reviewed!" as a defense of his work from AI audits, I'd encourage you to look at his two big papers in the Quarterly Journal of Economics from 2016 and 2018.
Both got a rapid single round of R&R within 2-3 months of submission date and both were accepted by the journal within 2-3 months of that revision. Both were up on the journal's website in under a year.
Now I don't have any special insight to what the referees on these two papers said. But I do know that this is an extremely rapid submission-to-print timeframe for a complex empirical article at a top economics journal. And it's especially rapid considering that people were pointing out problems with the working paper versions of both while they were still under review, and most of those problems remain unaddressed.
For comparison, the Auten-Splinter income series (which is the most direct competitor to PSZ's 2018 QJE piece) took FIVE YEARS of rigorous peer review back-and-forth between its initial submission in 2018 and its acceptance in 2023 at JPE. @VincentGeloso and I first released our critique of a different part of Piketty-Saez's data series in 2017. It took multiple rounds of rigorous review and back-and-forth edits on every section before we landed at EJ in 2022. I suspect that what these papers experienced is much more in line with the norms of a heavy and rigorous peer review at an A-list journal.
IOW, I consider it highly unlikely that the <1 year rapid turnaround that SZ and PSZ got twice in a row at QJE is anything remotely common at that or any other top journal.
Carl Benedikt Frey retweeted
Carl Benedikt Frey considers why some societies are so much more suspicious of the technology than others. project-syndicate.org/commen…
Carl Benedikt Frey retweeted
I'm thrilled to be launching Econvitals today. It's been the main way in which I have been tracking and understanding the economy over the last several months. And I decided I might as well share it with all of you--for free.
Has a combo of website, email lists & social media.
Carl Benedikt Frey retweeted
Thor Berger @bergerthor has written a great paper on the effects of the diffusion of tractors in the United States, with a very nice thread that describes the analysis. The widespread adoption of tractors led to increased human capital investment both because children of farmers left agriculture. On the other hand, those areas that adapted most quickly exhibited the largest graduation effects, suggesting human capital investment was able to generate complementaries between farmers and tractors as the technology of agricultural production evolved. This possibility, of course, has contemporary resonance.
Carl Benedikt Frey retweeted
While a plausible claim, I think this claim is missing some important context
“The idea that Mark Zuckerberg was sitting in his Harvard dorm room and would have been deterred from starting Meta if the capital gains tax rate was slightly closer to the ordinary income tax rate strikes me as categorically absurd,” argues @NatashaRSarin, professor of law at Yale Law School.
Carl Benedikt Frey retweeted
Institutions that once gave Europe a comparative advantage in catching up with America have become a constraint now that Europe must compete with the US at the technological frontier, notes @carlbfrey. bit.ly/46ShgiG
Carl Benedikt Frey retweeted
What puzzles me about French economists like Gabriel, as well as Emmanuel and Tomas, is the zeal they bring in trying to convince Americans that they have to implement a French model that is focused on redistribution rather than growth, because every indication we have suggests that the focus on redistribution has cost France, and some other countries in Europe, like my own native Belgium, very dearly so far. In fact, the French economy looks like it is about to enter a serious fiscal crisis because its redistribution model is no longer sustainable.
Californians ought to celebrate the fact that wealth is being created in California, and not somewhere else in the U.S. or in the rest of the world. Americans, and Californians, for that matter, should continue to do what they do best, which is focus on growing the pie. It would be a mistake to adopt a model that is fundamentally broken.
What the French should do instead is try to copy the model that made California an engine of innovation. In that case, smart, highly educated French kids wouldn't come to the U.S. to start companies. They would do so in France. If they had done that, they would have more than one AI startup company.
More defence spending (though needed) or an "Airbus for AI" is not going to restart European growth. My five cents in @ProSyn on what Europe needs.
Link in comments.



























