Arpit Gupta

I'm teaching a new course at Stern on AI in Finance and opening it up! Syllabus/slides on Github: github.com/arpitrage/ai-i… Weekly summaries on Substack arpitrage.substack.com/p/1-three-rule… First post is on Amdahl's Law, Jevons' Paradox, and why finance was slow to learn the bitter lesson

One way to address the natural concerns about funding Princeton's schools would be for Princeton University to pay property taxes to support the education in the college town, which goes disproportionately towards educating the children of Princeton's own faculty and students

Jevons paradox all over the place

Week four of AI in Finance: • The economics of market intelligence • GPT can read a 10-K, but so can everyone else, so where does alpha go? • How to simulate your customers arpitrage.substack.com/p/4-the-end-of…

Week three of the AI in Finance course: • The economics of credit scoring from ledgers to ML and digital footprints • How AI lets banks lend to strangers • Why better models create new risks arpitrage.substack.com/p/3-the-cause-…

I'm teaching a new course at Stern on AI in Finance and opening it up! Syllabus/slides on Github: github.com/arpitrage/ai-i… Weekly summaries on Substack arpitrage.substack.com/p/1-three-rule… First post is on Amdahl's Law, Jevons' Paradox, and why finance was slow to learn the bitter lesson

Check out the new state land-use reform tracker from @FurmanCenterNYU. It's a great resource! furmancenter.org/data-tool/land…

The Jones Act is just a really tight pair of shoes that you take off during national security emergencies to make your feet stop hurting when the time is right

Why do major AI models tell left-wing voters in Japan to vote for the communist party? My new research paper led by Sho Miyazaki. In 2026, voters across the world will be asking AI to help them vote. How will the AI respond? We study this question in Japan, which recently held a snap election. When voters provide policy positions, we find that the models rely heavily on this information—and in Japan, the models heavily recommend the communist party in response to left-wing positions, even though the positions we provided are held by a range of other parties. Why are the AIs doing this? We’re not sure, but we have a theory: in Japan, the communist party operates a content-heavy, fully open website with a “newspaper” that is openly accessible for AI models. In contrast, many Japanese news outlets block AI models from accessing their content. The result: the Japanese Communist Party website is one of the most-cited “news sources” in our study. This pattern of recommending the JCP is consistent across many models, including the most recent frontier models. There’s much more work to do here, but we think our paper suggests two main takeaways: AI models should be more careful about what sources they consider news, maybe especially in non-US contexts where the model companies may hold less policy expertise Parties and news sources that want to influence AI recommendations should think twice about excluding their content from AI. To paraphrase @tylercowen, when it comes to elections and voting, journalists may want to “write for the AI”! Governments may want to consider policies that allow this content to be used for voting recommendations but not for other AI model use cases. Looking forward to everyone’s feedback as we prepare to submit this paper and turn to studying US voting recommendations in advance of November’s midterms. Check out the full paper below.

@grok @tyler__palmer You’re wrong @grok!

I analyzed the 5Y chart: started ~$70-80 in 2021, dipped hard to ~$10 in 2022, hovered low ~$10-40 through 2023 into early 2024, then exploded—up to ~$320 by end 2024, peaked over $700 in 2025, now around $460 with recent volatility. Classic late bloomer AI-adtech play. That's AppLovin (APP). Spot on?

Guess this stock

@omzidar @paulnovosad Also worth noting Princeton public school enrollment is stable if not declining. They use a "multiples" approach to back out likely enrollment from housing units, and not clear how much declining fertility impacts these estimates.

@omzidar @paulnovosad This particular PILOT skips the public schools (not included in the negotiation), but other local projects like this one direct the money to public schools directly planetprinceton.com/2025/05/28/pri…

Im generally pro more housing, including in Princeton, but I think it needs to be done in conjunction with other fiscal and service expansions. there are a few issues this Washington post story missed - in my view, historic preservation isn’t the main point. One issue is that many of the new housing units get pilots so they don’t contribute property taxes for something like 10+ years. Adding ~ 1000 people to a small school zone and not increasing any tax revenue to pay for it causes genuine congestion problems. If the schools and staff got twice as big when the local pop doubles, that’d be fantastic. but they don’t, and people reasonably don’t like that the drop off line takes forever bc it’s not big enough to drop off 2x as many kids, their class sizes have ballooned, kids play areas don’t work bc the space wasn’t designed for that many kids, and most importantly, school quality falls as the educational inputs don’t scale with population. Practically it’s hard to expand schools and investments for young people in a town with a large share of elderly people, several of whom tend not to support higher property taxes or increased school spending. So adding more housing beyond the recent increases (which I think increased the overall population by like ten percent again with little new tax revenue) is more controversial for some of these reasons. Moreover, there are also other pressing fiscal issues for the town to deal with - it’s spent a lot on various projects and it’s unclear how much is paid for and how much prop taxes in a high prop tax place will have to go up already to cover deficits - so adding even more people without getting more tax revenue (bc of pilots and the way these things are implemented) creates additional reasons for opposition.

China has poor capital discipline — it issues a ton of equity, which dilutes investors, but supercharges investment

Leopold Aschenbrenner predicted in June 2024 that we would get a dramatic improvement in AI capabilities around the turn of 2026 due to the switch from chatbots to agents, which he thought would unlock a new set of AI capabilities Which is basically exactly what happened?

@paulnovosad My guess has been that Trump feels "they started it" with the assassination attempt on him

Will this new kind of warfare come back and bite us? Nah, I'm sure it won't

@razibkhan I'm from the red area; can digest milk easily

to be clear, this is milk sugar. the fat and protein are still digestable. additionally, microbiome shifts probabably allow for some digestion too even without lactase persistence

Breaking: @pleiasfr and @nvidia release the first open synthetic dataset for personas in Europe: Nemotron-Personas-France. 1M synthetic French persons, with rich imaginary lives grounded on (complex) demographic distribution.

AI really can help education: Randomized controlled experiment on high school students found a GPT-4o powered tutor that personalized problems for students raised final test scores by .15 SD, "equivalent to as much as six to nine months of additional schooling by some estimates"

I spend way too much time on social media debunking "economic slop" promulgated by lawyers pretending to be economists, so I built Show Me the Model: a tool that uses AI to check whether the economic reasoning in an essay actually holds up. showmethemodel.io Give it a URL or paste some plain text, and the tool flags hidden assumptions, internal inconsistencies, and other problem areas, and tells you how a real economist would think through the issue. Right now, it has 4 "personas:" macro, trade, IO/price theory, and labor. The tool first figures out which persona is right for the job, and then uses a parallelized prompt scaffold specific to that persona to process the source text. Here are some example outputs based on some essays that triggered me hard: Citrini Research's viral essay on how AI could trigger a self-reinforcing financial crisis rivaling the GFC: showmethemodel.io/#/results/2ez3… American Compass on the harms of trade deficits: showmethemodel.io/#/results/kOvt… @oren_cass on why Built-to-Rent should be banned: showmethemodel.io/#/results/OXjr… American Compass on the "China Shock:" showmethemodel.io/#/results/dJM7… @michaelxpettis on why China's trade surplus reduces global output: showmethemodel.io/#/results/C8OT… Try it yourself at showmethemodel.io. You'll need to bring your own API key (OpenAI or Anthropic), and a typical analysis costs $0.50–$1.50. It's super preliminary and will probably break on you. I'd love feedback about both the functionality as well as the quality of the output.

@weakinstrument I sympathize, in that it's not how they intended to be read! I get that

@weakinstrument True, and I think also to be fair it's not that we have any proposals to tax the rich alone. But on the margin we have proposals to tax the rich more and rebate to the rest less. I do think that's intellectually downstream of their tax incidence framing

Ironically, the Piketty-Zucman view that the hyperrich have all the money has led to calls for lower taxes, as many Democrats now say taxes can go down for most people and we can raise all the money we need on the super rich