Adam Altmejd

AI is already improving itself.

The War Trump Chooses That wasn't Trump against Zelens'kyi. It was Americans against reason

The attempted mugging of a visiting president was about the world war that Musk, Trump, and Vance have chosen. If we attend to what Vance and Trump said yesterday, we can work our way to the unreason of American policy, and to the chaos that will follow.
snyder.substack.com/p/the-war-tr...

Stort grattis!!

We use a new machine learning technique to identify the most cost-effective policies to boost immunization demand in India. We evaluated 75 combos of reminders, incentives, and local community ambassadors. The most effective increased immunization by 44%. https://buff.ly/4j0v3IF

DN Debatt. ”Vi ändrar oss – skrota betyget F” DN Debatt. Kristdemokraternas Jakob Forssmed och Mathias Bengtsson om skolan – vill också slopa den generella gränsen för gymnasiebehörighet.

Mycket bra. Betyder det att det finns en majoritet i riksdagen för Henreksons förslag nu? www.dn.se/debatt/vi-an...

Utmärkt förslag.

Bookmarking :)

I just got the kids down for naptime so I wanted to take a second to point out two key problems (and a few areas of agreement) I see with this morning’s @emmagoldberg.bsky.social @nytimes.com article about effective altruism: 🧵
bsky.app/profile/nyt...

FHM: Fler måste vaccinera sig Smittspridningen av influensa och covid-19 väntas öka under vintern. Alldeles för få personer vaccinerat sig, enligt Folkhälsomyndigheten (

Hade jättegärna vaccinerat mig men jag får inte för jag bor i Stockholm 🙃 www.dn.se/direkt/2024-...

Tried it for the the first time a few weeks ago. The speed, clean syntax, ability to typeset tables directly from csvs, don’t think I’ll be able to go back.

Jag fattar verkligen inte. Han pratar om hur AI ska hitta cancer men nuvarande lagrum tillåter redan forskare att samla in data på hela befolkningen från en massa olika myndigheter, träna en AI-modell och sen publicera så vårdpersonal kan använda den.

That was fun. R-datatable is a bit more concise :).

Carl-Henric Svanberg, AI-kommissionens ordförande, om vikten av att satsa på AI: ”Sverige har en dataskatt som många andra inte har” - Ekots lördagsintervju AI-kommissionens ordförande Carl-Henrik Svanberg vill tillåta delning av personliga hälsodata mellan myndigheter, något som skulle kunna göra Sverige världsledande ...

Svanberg från AI-kommissionen motiverar OSL-ändring för datadelning mellan myndigheter med att det ska underlätta forskning. Har kommissionen missat att det redan är tillåtet? (11kap 3§) sverigesradio.se/avsnitt/carl...

Was really surprised by that framing. Isn’t 60-77 a really narrow range given the plurality of both opinions and methods?

Ooh thank you!

Looking very much forward to presenting my research on the inheritance of fields of study at the @tinbergeninstitute.bsky.social tomorrow!

I’ll be giving a new Econ msc-level course on applied methods/data science/AI tools starting this spring. Please send suggestions of topics to cover and best practices. (In addition to the awesome courses by @gmcd.bsky.social and @mattblackwell.bsky.social, of course!) #econsky

Was just gonna say the same, Dutch data is a mark of quality!

How do I sign?

🔗 [8/8] This study is part of our COVID-19 research program SWECOV (see swecov.se). For a more in-depth understanding, I invite you to read our full article: www.pnas.org/doi/10.1073/.... And please do share your comments and thoughts!

🤝 [7/8] By taking a bird's-eye view and applying uniform methods to gauge the many dimensions of inequality, our study provides fresh perspectives that complement the many specialized studies on the pandemic's impact.

🔎 [6/8] This strong persistence surprised us. It suggests a structure of inequality so rigid that even a global, unexpected, crisis leaves it unaltered. These findings underscore the need for pandemic policies to take inequality seriously.

📉 [5/8] (2) Interestingly, the pandemic didn't significantly reshape the structure of social inequalities. The relative positions of different social groups remained strikingly consistent with pre-pandemic patterns. For 2020, only 2/72 markers fall outside the 10%-change cone!

🌍 [4/8] Our paper reports two central findings: (1) Disadvantaged groups—those with lower incomes, education, and non-European immigrants—suffered more, both from COVID-19 and from wider health issues and economic strain. The graph reports risks relative to the population mean.

📈 [3/8] Life during the pandemic got worse in almost every way. We observed an overall increase in negative events, deviating from pre-pandemic trends. This uptick underscores the pandemic's sweeping impact on societal wellbeing.

📊 [2/8] Using comprehensive Swedish population data, we analyzed the relative risks of nine adverse events before and during the pandemic, across different social groups by gender, education, income, and birth region.

🧵 [1/8] Our latest research (with Olof Östergren, Evelina Björkegren, and Torsten Persson) published in PNAS sheds light on the disproportionate impact of the COVID-19 pandemic on already disadvantaged groups. Here's a thread summarizing our findings. 📉📈 #COVID19

Trying to answer an ethics review about why my data collection does not process personal data, but unsure.

GDPR says IPs are PII. But web sites (servers) use IPs for content delivery. This means all sites, routers, etc process PII? Can this really be true?

Is there a #gdpr-sky community yet :)?

As if she expected to win the Nobel, Claudia Goldin just published a new NBER working paper entitled "Why women won". www.nber.org/papers/w3176...

Grattis!! Så välförtjänt!