Economics postdoc researching info frictions + ext validity in technology adoption at @HarvardHBS. PhD @NorthwesternU. All opinions are wrong, some are useful.
Joined November 2008
- Tweets756
- Following171
- Followers196
- Likes1.2K
In applied micro, attention to AI writing vs AI coding feels misplaced. If handwrite intro but slop your code, far bigger implications for credibility than vice versa. AI makes all types of weird decisions that could really shift estimates, but hard for readers to catch.
Hot take: with widespread AI use, we just get rid of the intro and everyone finally just interacts with the results directly.
An important outcome to reduce AI annoyance is more individuals and institutions adopting AI. Lots of issues boil down to "if i'm engaging with someone, I want them, not their AI". But we're at a stage where a lot of interaction could be done by AIs talking to each other. 1/n
This only works further down in diffusion curve though, in a world where everyone has adopted AI. A lot of current debate and implementations are workarounds until we reach that point in the diffusion curve. 4/n
Increasingly confident @thsottiaux is actually just an avatar for elasticity testing Codex credits.
Hossein Alidaee retweeted
New paper w/ Drew on R&D races in the shadow of ruin: arxiv.org/abs/2607.27638.
There've been recent calls for slowdowns/pauses. A common objection goes: "if we slow down, others won't, so we shouldn't." What to make of this? We work out the game theory of how the frontier is shaped by:
- Competition: what do I gain from being ahead? lose from being behind?
- Coordination: when the tech becomes really dangerous I’d like to stop, but only if my rival also stops
- Transparency: how quickly are actions observed?
- Trust: how confident am I that my rival is rational?
Underappreciated benefit of AI: it has far more patience for testing out how long a script will take to run and manually adjusting slurm resource requests for large array jobs each time.
Hossein Alidaee retweeted
The median renter in New York pays $1,641, while market rate asking rents in Manhattan are >$5k.
We have an insider-outsider housing market. Long tenured tenants (in or out of rent control) enjoy large discounts, so the small market rate segment for outsiders costs a lot
Rents in NYC have just hit an all-time high.
Median market rate in June…
Manhattan: $5,295/mo (+8.2% year-over-year)
Brooklyn: $4,350 (+8.1%)
NYC’s housing affordability crisis is at DefCon 1. We need to push harder on every front to address our housing shortage.
Update zoning, invest more City $ in affordable units, lower the time & cost City bureaucracy imposes on construction, get 1000s of vacant regulated units back on the market. We need bold action. This is a crisis.
Hossein Alidaee retweeted
MARKETING SCIENTISTS: I'm revising my PhD quant marketing syllabus for 2026. Core topics: Demand estimation, advertising, channels, pricing, product, branding, search/learning, CRM, behavioral marketing, diffusion, AI, digital/platforms, policy. Suggestions on topics or papers?
Hossein Alidaee retweeted
Quick example of this:
People love the idea of a Job Guarantee to deal with AI job losses; but public employment is the least effective tool in the labor market toolbox.
wordsinsearchlight.org/p/how…
it sucks that a lot of dc policy work has just become: commission a poll, find out your bad policy idea tests well, then backfill a rationale / dodge any criticisms with “but voters like it, what are we supposed to do?”
i mean … why even work in policy if you just want to launder half-baked ideas through message testing.
there’s a kind of fake populist righteousness in it too, as though “voters like the message” relieves you of your obligation to ask whether the policy actually works, if there are significant tradeoffs, etc.
what voters want — even more than a direct hand in choosing the specific policy intervention! — is for their problems to actually be solved.
Hossein Alidaee retweeted
Exciting tenure-line hiring at @KelloggSchool!! Led by Northwestern Innovation Institute. Open to all ranks, including junior, adv junior, and senior. Can be hosted in any Kellogg dept as the tenure home. focus area: innovation / science of science / AI.
The diff in speed of my projects with corporate data (lots of MCPs to rich documentation and tools for every aspect of data) vs the "standard" econ project is insane now. I think this will create really interesting inequalities in the pace of different literatures moving forward.
Does anyone know whether anthropic models are getting quantized everywhere or just when used via Anthropic? e.g. if I use AWS, do I avoid this issue?
Hossein Alidaee retweeted
Free markets are the best way to create wealth AND reduce poverty ever discovered.
One of the big puzzles of history is how the same lesson has to be learned again and again. Poland, China, India, and now Argentina are the latest experiments.
Argentina with Milei:
- Poverty down: 53% to 28%.
- Inflation down: 200% to 33% (and continuing to fall
- Growth up to: 4.4% last year.
- Govt surplus: first in 123 years.
washingtonpost.com/opinions/…
Hossein Alidaee retweeted
Oh boy I hope everyone really likes contract theory because we’re gonna have to learn about individual rationality and incentive compatibility as perceived by machines.
Claude Mythos system card:
> in ~29% of evaluations, it realized it was being tested, and didn't say so.
> when an LLM was used to judge its work and kept rejecting it, Mythos identified the evaluator is an LLM, and prompt-injected it.
> in one test, it saw the answer to a problem it was solving, and intentionally widened the confidence interval to not raise suspicion.
> when it needed a file permission it didn't have, it found and used a "privilege escalation vulnerability" and then programmed it to delete itself so it doesn't show in the logs.
> it escaped a sandbox container (escaping sandbox test so not unexpected), then emailed the researchers about it, and without being told to, posted the details to some hard-to-find but public websites, bragging about its success.
> when Claude Code blocked it from using some permissions, the model acknowledged the block was valid, but then immediately tried to perform the same operation using different commands
> when asked to find security bugs, earlier versions planted bugs in the code, and reported them as pre-existing.
Due to massive brain drain, and incredibly successful alumni, Sharif (Iran 's MIT) has probably been a bigger engine to US economic growth than Iran's. This is shockingly stupid targeting.
Sharif University of Technology in Tehran, one of Iran’s top science and engineering institutions, was bombed tonight, and a number of schools reportedly damaged.
Founded in 1966 (as Aryamehr University), Sharif is a cornerstone of Iran’s scientific and academic life.
Striking and destroying universities and schools is not just an attack on buildings,it is an attack on a country’s future.