The post William Nordhaus, RIP appeared first on Marginal REVOLUTION.

       

Comments

• “As an economist and as a university citizen, Bill Nordhaus ... by MikeP
• He was great indeed. I liked his paper on the singularity : ... by mento

Related Stories

• And so it begins…
• My Barcelona talk
• The pass-through of retail crime

 

Researchers, I just uploaded a prominent QJE paper on climate change economics to @RefineDotInk and I loaded my critique of the paper. I asked the AI to offer an impartial verdict. I will make this report public. A new journal could collect these into “Special Issues”.

That is from Matthew E. Kahn.

The post And so it begins… appeared first on Marginal REVOLUTION.

       

Comments

• In reply to Steven Kopits. OTOH to see why many people ignore ... by Wayfaring Stranger
• In reply to Cancelled for wongthink. If you wanted to scare ... by Steven Kopits
• “The authors find that a permanent 1°C warming reduces world ... by Steven Kopits
• Climate change makes me think of the scene from Trains Planes ... by Cancelled for wongthink
• Turgid. by Steven Kopits
• Interesting , Of course people adapt. People have adapted to ... by mento

Related Stories

• Can Artificial Intelligence Forecast Treatment Effects in Economics?
• How is AI shaping economic research?
• How Tyler Cowen and Alex Tabarrok Have Changed Their Views on AI and Where They Disagree

 

2. World Bank African urbanization report.  Building out African cities in a decent way could end up being the task of our coming era.

3. Summary of The Commanding Heights.

4. “Meanwhile in Armenia…”

5. “Female psychopathology has worsened despite increased resources and opportunities, suggesting it does not entirely reflect absolute deprivation, but rather, is a competitive adaptation activated by heightened social competition.”

6. One account of what is going on in Poland.  I am not endorsing that (or any other) interpretation, but if nothing else it shows you how interpretation of these events can make your head spin.  Grok has a simpler answer, again not endorsing.

The post Saturday assorted links appeared first on Marginal REVOLUTION.

       

Comments

• In reply to Dismalist. Great insight, Dismalist: Kissinger ... by Kaz Stanczak
• In reply to Kaz Stanczak. Stephen Kotkin, author of ... by Dismalist
• In reply to Mark Bahner. Mark, think: Germany's relative ... by Kaz Stanczak
• In reply to B. Theoretical hypothesizing — or hypothetical ... by mkt42
• In reply to Bob. Your first sentence is worthy of a Karl ... by Dismalist
• Plus 10 more...

 

The post My Barcelona talk appeared first on Marginal REVOLUTION.

       

Comments

• Youve been down on Catalan nationalism. I hope this comes up.in ... by Hoosier
• In reply to Anonymous. Heh! Seriously, if we go by the order ... by mkt42
• Congratulations on winning the prize! Try Embat. Batea is ... by AG
• Congratulations on the prize. Well deserved. Was it for the ... by Anonymous

Related Stories

• The pass-through of retail crime
• Better Incentives, More Kidneys, More Lives
• Can Artificial Intelligence Forecast Treatment Effects in Economics?

 

This paper shows retailers increase prices in response to organized retail crime. We match store-level crime data to scanner data from the universe of transactions for cannabis retailers in Washington state. Using quasi-experimental variation from robberies and burglaries, we find a 1.7–1.8 percent price increase at victimized stores and nearby competitors. This rise is not driven by short-to-medium-term demand changes but is consistent with crime-related cost shocks. Effects are larger for independent stores and less concentrated markets. We estimate crime imposes a 1 percent “hidden” unit tax on affected stores, implying $30.7 million in additional social costs, primarily borne by consumers.

That is from a newly published paper, AEJ: Economic Policy, by Carl Hase and Johannes Kasinger.

The post The pass-through of retail crime appeared first on Marginal REVOLUTION.

       

Comments

• In reply to rayward. Retail theft is nothing compared to the ... by rayward
• In reply to steve. Steve, Not sure what you mean by ... by Vivian Darkbloom
• 🙂 Yeah, its kind of outmoded but you see it everywhere. ... by Bubbles
• In reply to Kaz Stanczak. Its worse selling discount on line ... by Bubbles
• In reply to Gilligan. Right. They should switch to selling ... by kwazii
• Plus 10 more...

Related Stories

• Inequality in Equal Land
• Better Incentives, More Kidneys, More Lives
• Are there too many misdemeanor convictions?

 

Education is widely recognized as the primary driver of human capital accumulation. However, policies designed to promote educational attainment often overlook students’ endogenous responses. This paper provides empirical evidence of the causal impact of relaxed academic standards on student effort and achievement. We find that academic leniency results in mechanical GPA increases, but significant reductions in effort among lower ability students based on lower attendance and lower numerical grades. These heterogeneous effects compound and exacerbate gaps in engagement and achievement throughout high school, ultimately worsening deficits in human capital accumulation as measured by ACT scores.

That is from a newly published paper, in AEJ: Economic Policy, by A. Brooks Bowden, Viviana Rodriguez, and Zach Weingarten.

The post The unintended consequences of academic leniency appeared first on Marginal REVOLUTION.

       

Comments

• In reply to JP. There is third party observation — of the ... by Dismalist
• In reply to Thomas Pinchin' one out. which is basically ... by Wayfaring Stranger
• In reply to Gilligan. They do, there is inflation there too, ... by TMC
• In reply to Bill Allen. School is the primary cause of ... by Thomas Pinchin' one out
• In reply to Pipsterate. When you were involved with the ... by Thomas Pinchin' one out
• Plus 10 more...

Related Stories

• The pass-through of retail crime
• Are there too many misdemeanor convictions?
• Chronicle of Higher Education interviews me on AI and education

 

2. New and pretty brutal results on CTE and football.

3. Interview with Dan Wang.

4. Updated version of the Saez-Zucman audit.  Contestable and speculative assumptions are now separated from true defects.

5. Rosalind Arden interviews Greg Clark on social status.

6. GPTs vs. Claudes, in games.

7. The latest from Sam Altman.

The post Friday assorted links appeared first on Marginal REVOLUTION.

       

Comments

• 2. And going through school and not really learning anything ... by Mort Main
• In reply to Secrets. I've always thought the answer to “Why ... by C
• In reply to Mm. Yes, that is clear, but I'm still confused. ... by Ymxj
• In reply to RAD. While there is a point you have here, we ... by M
• In reply to mento. I think the key point for me, which is ... by M
• Plus 10 more...

 

“Organs are literally lost and damaged in transit every single week. The OPTN contractor is 15 times more likely to lose or damage an organ in transit than an airline is a suitcase.”

Organs are not GPS-tracked!

“In an era when consumers can precisely monitor a FedEx package or a DoorDash dinner delivery, there are no requirements to track shipments of organs in real time — or to assess how many may be damaged or lost in transit.

“If Amazon can figure out when your paper towels and your dog food is going to arrive within 20 to 30 minutes, it certainly should be reasonable that we ought to track lifesaving organs, which are in chronic shortage,” Axelrod said.

Here’s one more astounding statistics:

“Seventeen percent of kidneys are offered to at least one deceased person before they are transplanted….”

Did you get that? The tracking system for patients is so dysfunctional that 17% of kidneys are offered to patients who are already dead–thus creating delays and missed opportunities.

Fortunately, even then there were changes in the works. In 2019 President Trump issued Executive Order 13879 to improve the performance of OPOs with better incentives and to allow living donors to recover more of their costs. The government being what it is, it has taken years to implement some of the new rules and enforcement of the  OPO performance measures was deferred to 2026-2027. Nevertheless, there have been improvements. For example, many OPOs now GPS-track kidney transfers. And a new report estimates there have been significant gains from the changes.

• The Advancing American Kidney Health initiative was launched by President Donald Trump via Executive Order 13879 on July 10, 2019. In this Data Point, we estimate the range of increased growth in transplants that might be attributed to the policies launched through the Executive order and the economic value of that increased growth.
• The estimated impacts of the Executive Order (EO) range from a conservative 1,055 additional transplants annually to an upper estimate of 2,110 additional transplants. The range reflects variation in the counterfactuals used to compare with post-2019 transplant growth.
• Under the conservative-case scenario, the policy is estimated to be associated with approximately 23,214 additional kidney transplants between 2019 and 2040, generating about 147,410 life-years gained and 21,821 quality-adjusted life-years (QALYs) gained. Under the optimistic-case scenario, the policy is estimated to be associated with approximately 46,428 additional kidney transplants between 2019 and 2040, generating about 294,819 life-years gained and 43,643 quality-adjusted life-years (QALYs) gained.

The biggest and most consequential change would be to allow payment to living organ donors. The End Kidney Deaths Act, H.R. 2687, proposes a $50,000 refundable tax credit over five years for qualifying nondirected living kidney donors. The act has bipartisan support. Because dialysis is so expensive—and the federal government pays for much of it through Medicare’s special kidney-failure entitlement—the Act could save taxpayers money as well as saving lives.

Hat tip: The indefatigable Elaine Perlman.

The post Better Incentives, More Kidneys, More Lives appeared first on Marginal REVOLUTION.

       

Comments

• It’s great to see you highlight this issue and Jennifer ... by Doc b
• In reply to dave. Either way, the bullet points in Alex's ... by ShatGPT
• So the latest Resident Evil movie was a documentary? Probably ... by Crikey
• “The biggest and most consequential change would be to allow ... by Dick King
• In reply to rayward. Are do you purposely ignoring the ... by mikeInThe716
• Plus 10 more...

Related Stories

• The pass-through of retail crime
• Protecting Americans from Medicine
• William Nordhaus, RIP

 

Misdemeanors account for three-quarters of criminal cases filed annually in the United States, yet causal evidence on the impacts of misdemeanor conviction is scarce. We combine administrative and survey data to provide novel evidence of how misdemeanor conviction affects the socioeconomic outcomes of defendants and their families. Using data on nearly seven million defendants from 11 states, we characterize the types of people involved with the misdemeanor system. We estimate causal impacts of a conviction relative to a charge using seven county reforms that sharply reduced conviction rates. Regression discontinuity design estimates yield precise null average effects of the reforms on criminal and labor market outcomes. Defendants dismissed by the reform are unlikely to re-offend if dismissed, but also unlikely to be harmed by conviction—marginal defendants are older and more likely to have been charged with minor crimes than the average defendant. Heterogeneity analyses suggest that conviction harms young and first-time defendants, implying that interventions targeting dismissal toward these groups would be beneficial. Finally, our causal estimates imply a negative Marginal Value of Public Funds of misdemeanor conviction, even under plausible estimates of general deterrence effects.

That is from recent research by Nikhil Rao and James Reeves, note that Rao is this year on the job market from University of Michigan.

The post Are there too many misdemeanor convictions? appeared first on Marginal REVOLUTION.

       

Comments

• What is it about the conviction that “harms” the young men ... by Lizard Man
• In reply to TWBC. Have you ever gone out to a campsite in the ... by rsm
• In reply to rsm. What kind of wrong assumptions do you have ... by rsm
• In reply to Ted Bundy Lookalike. Having been to China in 1989 ... by rsm
• In reply to rsm. Sounds like a capitalist lie to me. YMMV. ... by TWBC
• Plus 10 more...

Related Stories

• The pass-through of retail crime
• Inequality in Equal Land
• The unintended consequences of academic leniency

 

Cowen: …Another proposal I’ve made is that schools should let people do one class, only one class, AI-only — no professor, except a professor grades at the end. The professor grades not answers to an exam but the workflow of how the student used AI to learn the Glorious Revolution, or the Civil War, or whatever the student learned. Eventually, another AI will grade your AI log. But let’s start today with a professor, and try this for one class. I’m not sure how it will go. But especially for small schools that don’t offer that many classes, a student now can create a class in literally everything.

If it’s successful, let them try two classes. I think the students doing this on their own are more likely to figure out how to blend AI and education than we would from a distance with some kind of centralized vision. Let them experiment.

Carlson: Students should be leading the change at colleges and universities?

Cowen: Absolutely. Students know more about AI than their professors do. Some of this comes from cheating, which is unfortunate. But the notion that the professors teach the students AI — I wish. It’s not always possible. You could imagine setting up a class where the students teach the professors AI, and we give the students credit for teaching. I’m half joking, but it’s not, in fact, a crazy idea. We need to experiment along a number of fronts. I’m quite convinced no one knows how to do this. It hasn’t been done before. So again, just one class. In general, schools hate this idea. They see it as opening the door to rendering some professors obsolete.

And this:

Carlson: There are a number of students who are deeply opposed to AI. What place do they have in this university that you’re thinking about?

Cowen: I would love to have different schools with different policies. If some universities try to ban AI altogether, that’s great. Let the competitive process operate. Personally, I would bet against those schools. For a long time, I’ve been worried that within a given tier of school — big state school, community college, Ivy League — there’s not enough diversity. AI policies will force a kind of diversity on us. So I’m all for no-AI schools, even though I would not go to one.

Recommended.

The post Chronicle of Higher Education interviews me on AI and education appeared first on Marginal REVOLUTION.

       

Comments

• In reply to Slocum. “most universities (and hopefully K-12 ... by Bubbles
• In reply to Mark C. Textbooks have their role but are ... by Joseph Ferguson
• In reply to Mandarin Learner. Try the Pimsleur audiophonic ... by Susanna
• “Students know more about AI than their professors do.” — ... by Dave
• In reply to Copilot-rsm. It seems to me that you've helped ... by Bill Allen
• Plus 10 more...

Related Stories

• And so it begins…
• The unintended consequences of academic leniency
• Can Artificial Intelligence Forecast Treatment Effects in Economics?

 

I use LLMs to forecast treatment effects of research studies pre-registered on the AEA RCT Registry and the Open Science Framework. I focus on studies whose results were first publicly released in 2026 and elicit forecasts from models whose training or knowledge cut-off dates were before the relevant set of papers was released. First, LLM models extract forecasting questions from the pre-registrations or pre-analysis plans. These questions elicit forecasted treatment effects; not all resolve in the subsequent papers, but those that do can be used to consider forecast accuracy. Forecasts were elicited for 889 treatment effects from 39 papers with results disclosed in January 2026 or later, or 673 from 33 papers after screening. Forecasts were elicited in raw units but I focus on those that could be standardized and therefore compared across papers. Currently, the median absolute forecast error is 0.054 SD for the best model, the average absolute error is 0.093 SD, and the MSE is 0.0218. The AI forecasts perform substantially better than benchmarks forecasting 0 effect or forecasting the mean effect of other papers’ results in that time period. Overall, these results suggest considerable scope for forecasts to be of value in informing policy and research.

From the excellent Eva Vivalt.  As I have been predicting, so much of economics (and science) is going to change, and radically.  Here is Eva’s thread on the paper.  Perhaps we should dispense with many of the more mundane inquiries, and as Eva notes: “But can we not shift to working on those projects that have particularly high uncertainty and upside potential?”

She is one of the best young economists in the world, and this stands some chance of being the most important paper of the year.

The post Can Artificial Intelligence Forecast Treatment Effects in Economics? appeared first on Marginal REVOLUTION.

       

Comments

• In reply to mento. https://econvitals.org/ Good site with a ... by TMC
• In reply to Stormy. Appreciate an informed opinion. Please ... by AI Guy
• In reply to Dismalist. Can anyone else write prompts to get ... by Gemini-rsm
• In reply to Dismalist. Gemini, after a few prompts: You are ... by Gemini-rsm
• In reply to rsm. I have no doubt that if you hand Galileo to ... by Dismalist
• Plus 10 more...

Related Stories

• And so it begins…
• How Tyler Cowen and Alex Tabarrok Have Changed Their Views on AI and Where They Disagree
• AI dominates human peer reviews

 

2. Video making the case that AI has been fine for jobs.

3. John Repetti reviews The Rouse at Jacobin.

4. Is TikTok now encouraging more optimistic content? (NYT)  Big if true, I am not currently active on the platform and do not have an opinion on the matter.  In any case worth a read.

5. Andy Masley recommends some lesser-known books.

6. New OpenAI dataset on how people use ChatGPT.

7. Piketty-Saez-Zucman biases found earlier by humans.  And an audit from Grok on whether the initial AI report is capturing the major problems.

The post Thursday assorted links appeared first on Marginal REVOLUTION.

       

Comments

• In reply to Gilligan. Prediction: by 2030 (2035 at the ... by cthulhu
• In reply to Gilligan. @Gilligan I see your point but then the ... by Torque CN Lewith
• In reply to M. M, of course, in theory at least, owners ... by Kaz Stanczak
• In reply to Gilligan. My understanding is that most teams ... by Matt
• In reply to M. (Apologies; error with the comment system) by M
• Plus 10 more...

 

Both became dramatically more confident in AI’s capabilities while remaining optimistic about its economic potential. Their differences concern how benefits arrive, how seriously to weight emerging dangers, and whether AI might possess consciousness. Cowen emphasizes institutions, bottlenecks, and geopolitical competition. Tabarrok emphasizes inexpensive intelligence, scientific discovery, and—especially in his later writing—concrete safety failures.

The dated passages below trace those changes across the public archive. Quotations are the authors’ own words, including excerpts from their columns; outside material they reproduce is identified separately.

Early views on unequal gains and abundance

Cowen’s early framework grew out of chess. On March 11, 2011, he observed that stronger computers required stronger human partners:

“The required education level — for the team’s ‘wage premium’ — is ratcheted up.”

His concern was already distributional: computers could increase productivity while changing which human skills commanded a premium. He also emphasized learning how to work with computers. What I learn from chess and computers.

Tabarrok’s April 12, 2016 argument was more explicitly reassuring:

“AI will be cheap and ownership will be widespread.”

He compared AI to an increase in the world’s supply of educated people: more intelligence produces useful ideas, and competition spreads the benefits to consumers. But his forecast of comprehensive substitution was cautious—AI becoming a perfect substitute for human intelligence could be “100 years or more” away. Will the proliferation of affordable AI decimate the middle class?

Neither position was absolute. Cowen warned in February 2017 that the Industrial Revolution involved decades of painful adjustment; historical progress did not guarantee quick wage gains. Tabarrok, in October 2019, praised the argument that mediocre automation could displace workers without creating enough additional wealth to compensate. See Cowen’s historical warning and Tabarrok’s Beware the Mediocre Robots!

Rising confidence in AI capabilities from 2020 to 2025

Cowen was already optimistic about language models in July 2020. After publishing a skeptical reader’s assessment of GPT-3, he added:

“I remain bullish, but it is always worth considering other opinions.”

That attribution matters: the skeptical arguments in the post were the reader’s, not Cowen’s. The case against the import of GPT-3, July 23, 2020.

Tabarrok’s shift toward near-term transformative AI became unmistakable by March 2, 2023:

“I never expected to witness the birth of aliens.”

Later that month, he endorsed Microsoft researchers’ description of GPT-4 as exhibiting an incomplete form of general intelligence: “They are correct.” This is a substantial acceleration from his 2016 outlook, although perfect substitution for humans and early general intelligence are different thresholds. See AGI is Coming and Baby AGI is Here, March 24, 2023.

Cowen reached an especially explicit formulation on April 16, 2025, discussing o3:

“I think it is AGI, seriously.”

But he immediately separated capability from adoption:

“It still will take us a long time to use it properly.”

His growing enthusiasm about the technology did not imply an equally dramatic forecast for immediate economic growth. o3 and AGI, is April 16th AGI day?

Economic transformation and the pace of adoption

On January 6, 2025, Tabarrok wrote:

“The intelligence revolution is going to be bigger, more impactful and more wrenching than the industrial revolution.”

This is his own concluding sentence, following a quotation from Sam Altman. The Intelligence Revolution.

Cowen’s February 23, 2025 response to rapid-takeoff expectations focused on a different constraint:

“Human bottlenecks become more important, the more productive is AI.”

Energy, regulation, organizational habits, and weak human collaborators can constrain the economic value of better models. He offered an explicitly tentative estimate of an additional half percentage point of annual growth—large when compounded, but far from an immediate economic explosion. Why I think AI take-off is relatively slow.

By September 3, 2026, Tabarrok was developing the distributional implications:

“A smaller share of a much larger economy can still mean more income for labor.”

He argued that even substantial displacement could leave aggregate labor income higher, or make compensation affordable. He explicitly acknowledged painful losses for particular workers. His high-growth calculations are scenarios, not forecasts. How Much Redistribution Will AI Require?

They also converge more than their rhetoric suggests. In May 2026, Cowen rejected mass unemployment while identifying transitional hiring problems. In August 2026, Tabarrok interpreted business-survey evidence as consistent with their shared position that AI had not yet broadly reduced employment. See Cowen, May 13, and Tabarrok, August 31.

Growing attention to safety and institutional limits

Cowen stated his central position on November 19, 2023:

“I view AI as more likely to lower than to raise net existential risks.”

His reasoning was comparative: the world already faces serious dangers, and stronger science could improve our defenses. My summary views on AI existential risk.

Tabarrok’s later writing places more weight on specific vulnerabilities. On April 8, 2026, he described:

“the worrying combination that very powerful AI’s are very vulnerable.”

He was discussing cybersecurity capabilities alongside agents’ susceptibility to manipulation. AI Risks.

On September 17, 2026, he sharpened the institutional objection to assuming that beneficial science will automatically offset dangerous capabilities:

“But faster science does not automatically mean faster defense.”

This can be read as a qualification of Cowen’s earlier optimism: defensive discoveries must pass through institutions that may move slowly. It also applies Cowen’s own bottleneck logic to safety. Pandemic Preparation is AI Safety.

Cowen nevertheless supports safeguards. In August 2026, he proposed government-overseen, industry-led safety audits. In September, he opposed restrictions on recursive improvement as impractical to enforce internationally. His position is therefore selective regulation combined with continued development. See his August 27 proposal and September 22 critique.

The September 2026 exchange about AI leaders

On September 16, 2026, Tabarrok rejected the claim that executives’ safety warnings were merely marketing or regulatory capture:

“Amodei actually believes what he’s saying.”

He did not thereby establish that the warnings or proposed policies were correct. His argument concerned sincerity and incentives. What Regulatory Capture Actually Looks Like.

Cowen replied the next day:

“I agree with Alex that the standard model of regulatory capture does not apply here”

But he emphasized competition for power: each executive sincerely thinks their own company should win. For Cowen, that complicates treating their warnings as disinterested policy advice. They agree that concern is sincere; they differ over which incentives best explain behavior. The correct model of AI CEO behavior, September 17, 2026.

Consciousness and Tabarrok’s warning about infant pain

On March 17, 2025, Tabarrok’s What Did We Learn From Torturing Babies? drew a moral lesson from medicine’s historical dismissal of infant pain. He described how theories about immature brains allowed physicians to discount babies’ observable distress and perform procedures without adequate anesthesia. His warning explicitly extended to judgments about animals and AI:

“If we must err, let us err on the side of compassion.”

Tabarrok’s position is stronger than mere openness to AI consciousness: he argues for moral caution under uncertainty. The analogy does not establish that AI suffers. It challenges confidence that a theory ruling out suffering is sufficient grounds to disregard possible evidence of it.

On October 14, 2025, he went further in his own commentary following an essay by Jack Clark:

“these systems are becoming self-aware”

He distinguished self-awareness from sentience, while warning against confidently dismissing the latter. His link accompanying that warning points back to the infant-pain essay, making the connection explicit. We Turned the Light On—and the AI Looked Back.

Cowen’s position is much more categorical. On August 31, 2026, he wrote:

“I am very much opposed to the view that the AIs are sentient, or might be sentient.”

He treats AI sentience as a category error, although he accepts that human-like descriptions can sometimes help predict AI behavior. Anthropomorphizing AI?

The disagreement concerns both what AI might experience and how to act when we cannot confidently know. Tabarrok’s historical analogy places greater weight on the moral cost of mistakenly denying consciousness; Cowen expresses much greater confidence that sentience is the wrong category to apply. These are their philosophical assessments, not established findings about AI consciousness.

The post How Tyler Cowen and Alex Tabarrok Have Changed Their Views on AI and Where They Disagree appeared first on Marginal REVOLUTION.

       

Comments

• In reply to Slocum. Two aspects I think you're missing are: ... by Mark Bahner
• I always thought that torturing animals was wrong, but not ... by Tim
• I published an in-depth analysis of Tyler's views on AI: ... by Ariel Krakowski
• In reply to mento. It’s slop. I find it so hard to read ... by Some Guy
• In reply to Max. I, for one, am glad this was published. It ... by Neurotic
• Plus 10 more...

Related Stories

• And so it begins…
• Can Artificial Intelligence Forecast Treatment Effects in Economics?
• How is AI shaping economic research?

 

We show that AI-generated article reviews are now better, by every measure we could construct, than reviews written by well-compensated human experts and judged to be of unusually high quality by journal editors and authors. We then discuss how this result may inform journal policy choices, with concerning downsides as well as potential paths to increase publication speed and quality; reduce adversarial interactions; improve the experience of editors, reviewers, authors, and readers; and advance science.

By Gary King and Steven Worthington, not even using what are now the very best models, rather based on Claude Opus 4.8.  The details of the dominance, as outlined in the paper, are striking.  It is not close.

On my Saez and Zucman post, I find it striking that some people are complaining all of a sudden.  Yet I have been offering, citing, and linking to commentary on published papers for twenty-four years now (and every single day).  Sometimes it might just be a casual offhand remark from me, whether positive or negative.  Typically this is simply taken in stride.  It is when you pull out the bazookas, and offer oodles of high-powered analytics, more astute than human referees, and combined with the ability to pursue follow-up inquiries, that people start getting worried.  Model that.

You can pick at individual criticisms from the AIs all you want, but the most damning observation is that the errors go in the ideological direction of the authors 97% of the time.  That is exactly what (the human) Phil Magness said years ago, based on an extensive canvassing of their work.  Basically we already knew that was true, and the AIs confirmed the human estimate.

On our side, we have taken steps, most of all the new MR archive, to make AI analysis of our work easier.  Soon you will be able to read on MR what the AIs think are our worst posts, respectively (Alex has prepared this).

In the meantime, I encourage the NSF, prize committees, universities, tenure committees, journals, and so on to do as many of their own AI-enabled audits as possible.  So far the world is not jumping at this opportunity.  You can model that too.

The post AI dominates human peer reviews appeared first on Marginal REVOLUTION.

       

Comments

• In reply to Vivian Darkbloom. It might be in general, though ... by M
• In reply to &. No, controls are needed when information *can ... by M
• In reply to Lord Action. LA: Virtually every product I buy ... by Wayfaring Stranger
• In reply to rsm. Reinhart Rogoff story was entertaining. by Dismalist
• In reply to Kaz Stanczak. The marginal revolution, man, the ... by Dismalist
• Plus 10 more...

Related Stories

• How is AI shaping economic research?
• Can Artificial Intelligence Forecast Treatment Effects in Economics?
• The pass-through of retail crime

 

Roman Mars has built one of podcasting’s great empires out of noticing things everyone else walks past. His show, 99% Invisible, has racked up over 500 million downloads telling stories about the unnoticed design choices that quietly shape how cities and objects work…

Tyler and Roman discuss street signs, why classic cars were so much more distinctive than today’s SUVs, Waymos, New Jersey jughandles, yellow streetlights, lane-stripe colors, why Houston keeps growing despite its lack of charm, whether any post-war American suburb is actually beautiful, the Chrysler Building’s rocky reception, brutalism done right and wrong, Central Park as an urban miracle, what Jane Jacobs got wrong, e-bike lanes, Tokyo’s bike-free transit model, the placelessness of global coffee shops, why the iPhone looks boring now, the ugliest country he’s visited, his love for Australia’s polymer notes, his soft spot for the dollar’s “locked in amber” weirdness, underground power lines, how to warn future civilizations about nuclear waste, why highway signs lie to you about speed limits, the principles of a great city flag, the Pocatello redesign saga, his own studio, why 99% Invisible works precisely because it’s audio-only, the most unusual habit behind his long productive career, and more.

Excerpt:

COWEN: Yes. One of my favorite topics, currency design. What’s your favorite currency in terms of how it looks?

MARS: Favorite? It’s like I love almost any of them with the US currency in dead last. I find things to love about them. I’ve done a story on Australian currency, which was one of the early polymer notes. I just like utility in currency. I like different sizes. I like different colors. I like the fact that they put on artists and people of note that are not founding fathers and presidents and stuff like this. I don’t know if I have a particular favorite, but I was entranced by the thoughtfulness and utility of Australian currency when I did a story about it. You could be blind and know what note you’re holding because of the size. They stand up pretty well, and they care about all kinds of things rather than just presidents and stuff.

COWEN: I loved Swiss currency pre-2021. They had Le Corbusier on, Giacometti, Honegger. In the 1980s, they had Euler on, the mathematician. Imagine putting Euler on your country’s currency.

MARS: That is a canvas that’s so available to us to represent more values than we have in the United States than just founding fathers and presidents and stuff. I think it’s just such a wasted opportunity.

COWEN: I like the dollar, though. I’m surprised you’re so negative. I like the weird esotericism: the eye, the pyramid. The design is bizarre. It announces to the world: “We’re crazy.”

MARS: There are things to love about it. I just think that its fussiness and filigree, we’re locked in, I think for important reasons, which is because the dollar is the most important currency in the world, to which all other currencies are pegged, and therefore you don’t change it lightly. You have to be very, very careful. It really locked it into amber in a way that is both beautiful to admire, but also really limits its utility as things move on, we get better things than linen-paper hybrids and especially different sizes and utility for people with different needs. I think that’s important. There’s definitely things to love about the dollar. The dollar has some charm.

COWEN: I like the arrogance of it also. A sense of we don’t need to make the denominations look so different. If you’re a foreigner, we expect you’ll come here a lot, you’ll master dollars. Maybe you use them in your home country as well. They don’t need to be different colors.

MARS: It is definitely a bold statement that you’re going to adapt to our way of being if you’re going to use the dollar, for sure.

Opinions throughout!

The post My design Conversation with Roman Mars appeared first on Marginal REVOLUTION.

       

Comments

• In reply to Gilligan. …and the old one was spectacular. by Larry Siegel
• In reply to David S. Indeed. I do notice that for each ... by ideatingfreak
• In reply to Wayfaring Stranger. That's correct about AZ and ... by mkt42
• In reply to Gilligan. You must have interesting Rorschach ... by Etalon d'Silomar
• Hadn't thought about the most beautiful street signage before ... by Trent
• Plus 10 more...