This essay makes three main points. First, even though US growth rates have been stable at roughly 2 percent per year for 150 years, it is distinctly possible that automating intelligence leads economic growth rates to accelerate. Second, this acceleration is likely to be slowed by the presence of "weak links." While we each have access to 100 million times more transistors on our desktop computer than people in the 1970s, we are not 100 million times more productive. Computers can invert matrices at lightning speed, but we humans must still decide what matrix to invert, what hypothesis to test, and so on. Accelerating economic growth requires the vast majority of the weak links to be automated away, which delays the large gains. Finally, even though weak links slow the benefits, they may actually speed up the risks. When a chain is only as strong as its weakest link, damaging one link in the chain can be very costly. A powerful AI that is superhuman at software engineering could be misused by a bad actor to do substantial harm by hacking the financial system or a virology lab.
This is a fairly standard economist’s take, I think. Jones hedges by writing
We begin by outlining two scenarios for the impact of artificial intelligence on the economy: one in which AI drastically accelerates economic growth and another in which AI is “business as usual.” Both scenarios are possible, and both scenarios have features that are informative about the future consequences of AI. . .This essay provides one perspective based on economic research, but the perspective is inherently speculative and will likely prove to be incorrect on many dimensions.
Colleges and universities are currently a brake on AI progress. As cultural institutions, responsible for forming the next generation of young people, the message is a lot of hand flapping about “needing to be workforce ready” and “don’t cheat” and “here are your required courses to graduate.” I’ve been focused for three years on the strangeness of higher ed’s preoccupation with “general education” and the sector’s commitment to staying the course while technology marches forward.
This is another difference between the 1990s and today. When I launched my web site in 1994, the academy was way ahead of the private sector with respect to using the Internet.
Schools teach algebra, history, biology, and standardized test preparation. Parents want AI literacy, mindset, and entrepreneurial thinking.
…I asked students, teachers, and professionals whether a college degree will become less important in an AI world. 79% said yes.
…The Socratic method is the pedagogy for the AI age.
…The ban-on-AI approach is the ban-on-calculator approach. It failed then and it will fail now. Write policy that mandates AI literacy, not AI prohibition.
Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen write,
We do not see widespread, economy-wide job displacement associated with AI.
However, young workers in AI-exposed occupations are increasingly falling behind their less-exposed peers. Employment among workers ages 22–25 in highly AI-exposed occupations now stands about 19% below where it would be if it had kept pace with employment among similarly aged workers in less-exposed occupations.
Pointer from Tyler Cowen
substacks referenced above: @
@
That's very interesting about where the Internet arose versus AI. The Internet came out of a collaboration between the DoD and the universities, and the web itself came out of CERN, a research center highly connected with the academy. Google itself is a private company, but its core technology for web searching is the Page rank, which came out of work Larry Page did at Stanford.
I learned about the Internet because I attended an event at Furman University as a kid and encountered a professor excitedly talking about Gopher, a predecessor of the web. I learned more about the web as an undergraduate CS student because Clemson used Internet networking for all of its machines, and because Prof. Harold Grossman encouraged me and the other students to learn about the web from both sides--both from installing a web browser on our personal machines to access our homework assignments, and in my case by installing a web server for others to use. The professors were part of the revolution and were teaching their students to use it so we'd be prepared for the future.
AI is a different story. The breakthrough technology was the new LLMs based on the transformer architecture, but universities could not match Google's ability to run the trials and experiments necessary to get the new LLMs to work well. I'm not close enough to it to be sure, but it seems like the new AI technology predominantly started in private companies, especially Google. (Google then lost some of its top AI technologists as they left for other companies.)
Before the AI breakthrough, there was also the breakthrough in translation of natural languages (e.g. French to English). It's not my area, but a researcher in the area I spoke to was demoralized about the conferences because Google kept blowing everyone else away. I asked why, and they said it's because Google has a big database based on the web, and academics couldn't afford that big of a database or that big of a compute installation to crawl and use that much data.
You can see why the academy would react differently to AI than to the Internet. When you're a professor and you and your friends are part of giant DARPA grants to build out the Internet, you're naturally inclined to show it to students and to encourage them to also join that same party. As the years go by, all those people you brought in will still be there and will be part of your network.
On the other hand, if you're not at a party, yourself, then it's much less attractive to push other people into it. You'd prefer it all to just go away.
Hollis Robbins provides much useful and interesting information. And I generally respect her work enough to be bothered to question three of her claims, specifically about federal funding, market failure, and forming new generations.
Robbins moans a lot about federal funding changes threatening universities’ ability to keep the lights on. But when we look at the actual federal budget tables ( see table 15.1, function 502 discretionary Higher Education at about page 16: https://www.whitehouse.gov/wp-content/uploads/2026/04/ap_15-1_fy2027.pdf ), we can see the inundation of the credentialing cartel with borrowed money proceedds apace, with $30,400 million down the drain in FY25, another $30,533 million circling in FY26 and $36,350 million proposed for FY27. Positively treasonous if you ask me given the national debt. Institutions that can’t keep the lights on absent such massive handouts really maybe ought to turn their lights out. And who, with a straight face, can even suggest they would be missed? Can anyone think of a memorable novel produced by a US college graduate this century? A painting? I can’t even think of a movie but then again I don’t watch many. If we want culture maybe we really ought to be thinking about looking elsewhere.
Second, Robbins claims that universities were founded to address a market failure. To be pedantic, historically this is completely false. US universities emerged from religious competition and local boosterism rather than a governmental attempt to correct an economic inefficiency. While some founders, like the first president of Bowdoin College, declared their institutions were "founded and endowed for the common good," the financial reality often contradicted this. Early colleges relied heavily on tuition fees and private donations, including funds derived from slavery, to survive. They operated more like private enterprises than public utilities, with many selling credentials to anyone who could pay. Modern proponents of the market failure argument suggest that because education generates positive externalities—benefits to society such as an informed citizenry and technical innovation—the free market would underprovide it if left entirely to private transactions. One reads Adam Smith claiming state funding was justified because an educated populace promoted domestic tranquillity and improved societal judgment. Ha ha ha. Frederic Bastiat, on the other hand, argued that government intervention often creates social costs that outweigh the benefits of correcting market imperfections. They suggest that state involvement can lead to inefficiency, lethargy among educators, and a misalignment of resources, as seen in the modern critique of administrative bloat and the decoupling of costs from value. Bingo for Bastiat.
Finally, Robbins claims universities are “responsible for forming the next generation of young people.” Dr. Kling has also written about how formative his Swarthmore seminars were and how this experience should be part of the student experience at universities. One need not strain oneself to imagine countless effective alternatives that don’t rely upon massive government handouts. Indeed, Against the wisdom of a medieval book, The Imitation of Christ, is probably the best antidote to all the maladaptive formation going on in US universities. Pope Benedict XVI explicitly endorsed the book during an August 2006 General Audience, referring to it as that "golden text of spirituality." Yet, as far as I can see, it is not required reading at any US Catholic university. Which tells you something about nominally Catholic universities.
At any rate, I will not direct the following quotes to anyone other than myself for I stand most in need of reminding of them:
“Every man naturally desires knowledge; but what good is knowledge without fear of God? Indeed a humble rustic who serves God is better than a proud intellectual who neglects his soul…”
“Shun too great a desire for knowledge, for it in it there is much fretting and delusion. Intellectuals like to appear learned and called wise. Yet there are many things the knowledge of which does little or no good to the soul…”
“Neglect of things which are profitable and necessary and undue concern with those which are irrelevant and harmful, are great folly.”
And on and on.
One suspects that if more parents strived more towards the ideal of say a William Godwin, the demand for universities would subside even more. Godwin home schooled his daughter, Mary Shelley, the acclaimed author of Frankenstein, through a rigorous, intellectual home-schooling model that defied the gender norms of the early 19th century. He provided her with a broad curriculum including ancient and modern history, literature, mythology, and the Bible, alongside practical lessons in art, French, and Latin. He published her early literary works and he immersed her among many wise people. Similarly, Dr. Kling used to write about his proposals prioritizing personal education, looking to letters of recommendation and introductions from a faculty network as a pathway to placement, and the separation of instruction and accreditation, allowing an independent entity to handle grading and certification. The nation can no longer afford the status quo: there are many such alternatives that might indeed prove superior were people left alone to pursue and develop them. And AI can help.