Anna Minasyan retweeted
Continuing to āspotā and call out people for using AI to write is stupid. Itās like trying to stop the sea with your bare hands. And this isnāt even really about AI slop.
1/9
Anna Minasyan retweeted
If you think students can benefit from writing research papers, read this article! You can still assign papers!
Dartmouth profs are doing this very successfully. One approach is having students use Google Docs, where the version history proves human writing.
It works! 1/
Anna Minasyan retweeted
I think the discussion is misguided.
AFAIK, Refine
- will not be used as a referee, but
- as an AI check for conditionally accepted papers w. authors being able to use feedback for final revisions.
Sensible to do that pre-publication rather than someone doing it ex-post.
Anna Minasyan retweeted
Apparently @sanjaygreddy, whom I was favorably quoting recently on RCTs, was just laid off from the New School despite having tenureāno other economist was laid off, only he was, presumably in retaliation for the years heās spent criticizing their financial mismanagement.
Sanjay Reddy at the New School wrote this piece seven years agoāright after Abhijit Banerjee and Esther Duflo won the Nobel prize in economicsāand it articulates better than I could everything wrong with this type of RCT. Who needed a study to "prove" that jealous, controlling husbands from poor backgrounds in one of the poorest parts of India prefer their wives work with other women than with men? The poorest women in India? Or the guy whose tenure case this study supports?
Anna Minasyan retweeted
This new PNAS paper is so interesting!
The authors asked a fascinating question:Ā Do we need language to reason logically?
They combined brain imaging in healthy adults with tests of two rare patients who had extensive damage to the brain's language network and profound aphasia.
Both patients showed severe grammatical impairments, often performing at or near chance on language assessments.
Yet their logical reasoning remained remarkably intact.
They could infer hidden rules governing transformations of number sequences, solving 19/25 and 39/40 problems.
They also excelled at nonverbal matrix reasoning, scoring 25/30 and 26/30āwell above age-matched norms.
The brain-imaging results pointed in the same direction. Neither inductive nor deductive reasoning substantially recruited the language network.
The conclusion is powerful:Ā natural language is not the medium of formal logical thought.
Language allows us to communicate thoughts, but it is not necessarily the machinery that produces them.
We can express understanding through language. But linguistic competence, by itself, is not evidence of understanding.
I think it is crucial to keep this distinction in mind in an age in which fluent machines are increasingly mistaken for thinking machines.
*
Full paper in the first reply
Anna Minasyan retweeted
The old line of asking a philosopher "Are you going to stay in academia or go into the philosophy industry?" no longer feels like a joke.
I've joined @AnthropicAI to work on alignment and character. I'll still teach at @nyuniversity; I'm on leave from @UTAustin.
New CEPR Discussion Paper - DP21660
Promotion through Connections: Favors or Information?
Yann BramoullƩ @amseaixmars @CNRS, Kenan Huremovic @IMTLucca
ow.ly/54aw50ZhHB7
#CEPR_LE #EconTwitter
Anna Minasyan retweeted
I was writing about land reform in West Bengal last night and was curious if it had persistent effects on the ownership distribution.
So I did what anyone would do, I* wrote an academic paper on it
Turns out ā yes! 1/
Anna Minasyan retweeted
Using labor force surveys from 160 countries to build a new microdatabase on hours worked covering 97 percent of the world population to explore working hours determinants, from @amorygethin and Emmanuel Saez nber.org/papers/w34217
Anna Minasyan retweeted
The deal between Armenia and Azerbaijan could end one of the worldās most intractable conflicts and a regional detente. Whether that happens will be a test of American diplomacy and of the countries themselves economist.com/europe/2025/08ā¦
Anna Minasyan retweeted
Just published in @JPubEcon:
"Does the prospect of upward mobility undermine support for redistribution?"
By @donandrewmoore, @ReneChoudhari, & @aileenwu_
sciencedirect.com/science/arā¦
#econtwitter #publiceconomics
Anna Minasyan retweeted
Which is true?
1. None of his advisors understand percentage changes.
2. At least some know Trumpās spouting nonsense but feel they canāt tell him.
3. Some actually tell Trump his math makes no sense and he ignores them.
Amazing how Groningen city center is being constantly improved and made convenient for pedestrians and cyclers. When your city changes that often no need to move anywhere š
Het college heeft het inrichtingsplan voor Oosterstraat en Gelkingestraat vastgesteld. bit.ly/4k3V3my Beide straten worden voetgangersgebieden. De straten krijgen een nieuwe inrichting waarbij voetgangers alle ruimte krijgen. De #raad050 besluit in juli over het plan.
Anna Minasyan retweeted
This paper confirms my belief that figuring out how to effectively employ LLMs at scale in education is one of the most important research problems of the day (and no, the answer is not āreplace teachers with AIā).
Huge potential returns, and the worst case is already happening
This quoted post is unavailable.
Anna Minasyan retweeted
10/ Please help share our extended replication to stop the spread of this erroneous policy evaluation!
Read the new report here: econstor.eu/handle/10419/316ā¦
Anna Minasyan retweeted
šØ REPLICATION REPORT UPDATE: One year ago, a tweet by @JohnHolbein1 alerted me, @OlleFolke, and Joop Adema (@Jopieboy) to a paper with a shocking result about Swedenās law criminalizing the purchase of sex.š§µ
Anna Minasyan retweeted
A re-analysis of Ciacci's (2024) "Banning the purchase of sex increases cases of rape: Evidence from Sweden" Journal of Population Economics reveals major issues. A year ago, reproducers Adema, Folke, and Rickne found coding errors driving the paper's key results. Let's unpack
ā¬ļø








































