@justusvonsoos.bsky.social
BWL at EBS, prev. Schloss Neubeuern
McKinsey Firsthand | E-Fellows Scholar
McKinsey Firsthand | E-Fellows Scholar
Reposted
Oh, wow, Gemini 3 Pro has solved 9/48 of the crazy hard FrontierMath tasks. And that's not even the Deep Think variant.
Previous record was 6/48 by GPT 5/5.1/5 Pro.
Previous record was 6/48 by GPT 5/5.1/5 Pro.
Gemini 3 Pro set a new record on FrontierMath: 38% on Tiers 1–3 and 19% on Tier 4.
On the Epoch Capabilities Index (ECI), which combines multiple benchmarks, Gemini 3 Pro scored 154, up from GPT-5.1’s previous high score of 151.
On the Epoch Capabilities Index (ECI), which combines multiple benchmarks, Gemini 3 Pro scored 154, up from GPT-5.1’s previous high score of 151.
November 21, 2025 at 8:31 PM
Oh, wow, Gemini 3 Pro has solved 9/48 of the crazy hard FrontierMath tasks. And that's not even the Deep Think variant.
Previous record was 6/48 by GPT 5/5.1/5 Pro.
Previous record was 6/48 by GPT 5/5.1/5 Pro.
Reposted
Another DeepSeek moment? Moonshot AI, a Chinese lab, released its new (open source!) model K2 Thinking, outperforming OpenAI et al. on several benchmarks. I tested it with a question from an unpublished paper of mine. Out of 5 tries, Kimi, GPT-5 and Gemini 2.5 Pro each replied correctly 3 times!
November 8, 2025 at 2:59 PM
Another DeepSeek moment? Moonshot AI, a Chinese lab, released its new (open source!) model K2 Thinking, outperforming OpenAI et al. on several benchmarks. I tested it with a question from an unpublished paper of mine. Out of 5 tries, Kimi, GPT-5 and Gemini 2.5 Pro each replied correctly 3 times!
Reposted
peer reviewers when you try to sneak in irony
November 8, 2025 at 11:04 AM
peer reviewers when you try to sneak in irony
Reposted
literally the last 2 places i'd go to for answers
November 6, 2025 at 9:53 AM
literally the last 2 places i'd go to for answers
Reposted
Dieses Streitgespräch zwischen @clemensfuest.bsky.social und @suedekum.bsky.social in der @zeit.de sollte man in Vorlesungen und Proseminaren zur Theorie der Wirtschaftspolitik durchnehmen. Sehr gutes Lehrmaterial, for the good and the bad. Ein 🧵:
October 18, 2025 at 9:31 AM
Dieses Streitgespräch zwischen @clemensfuest.bsky.social und @suedekum.bsky.social in der @zeit.de sollte man in Vorlesungen und Proseminaren zur Theorie der Wirtschaftspolitik durchnehmen. Sehr gutes Lehrmaterial, for the good and the bad. Ein 🧵:
Reposted
Wie die Exzellenzinitiative Deutschlands Universitaeten zwar verbessert hat, aber ein deutsches Harvard nach wie vor nicht in Sicht ist (auf PDF Seite 78 ff):
cloud.3dissue.net/19519/19560/...
cloud.3dissue.net/19519/19560/...
cloud.3dissue.net
October 13, 2025 at 12:48 PM
Wie die Exzellenzinitiative Deutschlands Universitaeten zwar verbessert hat, aber ein deutsches Harvard nach wie vor nicht in Sicht ist (auf PDF Seite 78 ff):
cloud.3dissue.net/19519/19560/...
cloud.3dissue.net/19519/19560/...
Reposted
The other day a student asked me about the prevalence of insider trading in prediction markets. I now have an answer.
October 10, 2025 at 11:19 AM
The other day a student asked me about the prevalence of insider trading in prediction markets. I now have an answer.
Reposted
Remarkable observation by Janan Ganesh
September 20, 2025 at 4:13 PM
Remarkable observation by Janan Ganesh
Reposted
Do tech optimists have a point? Within standard economic growth models, AI could drive explosive growth through one of two mechanisms.
1) Labor Substitution
So far, it seems like capital and labor mostly complement each other, which limits the returns to additional capital given fixed labor.
1) Labor Substitution
So far, it seems like capital and labor mostly complement each other, which limits the returns to additional capital given fixed labor.
September 19, 2025 at 9:35 AM
Do tech optimists have a point? Within standard economic growth models, AI could drive explosive growth through one of two mechanisms.
1) Labor Substitution
So far, it seems like capital and labor mostly complement each other, which limits the returns to additional capital given fixed labor.
1) Labor Substitution
So far, it seems like capital and labor mostly complement each other, which limits the returns to additional capital given fixed labor.
Reposted
A cautiously optimistic result on AI and disinformation.
A week before 2024 UK elections 13% of all voters used AI to ask about political topics. A randomized trial found this may be good: using AI led to similar gains in true knowledge as doing web research, regardless of model & prompt used.
A week before 2024 UK elections 13% of all voters used AI to ask about political topics. A randomized trial found this may be good: using AI led to similar gains in true knowledge as doing web research, regardless of model & prompt used.
September 18, 2025 at 8:15 PM
A cautiously optimistic result on AI and disinformation.
A week before 2024 UK elections 13% of all voters used AI to ask about political topics. A randomized trial found this may be good: using AI led to similar gains in true knowledge as doing web research, regardless of model & prompt used.
A week before 2024 UK elections 13% of all voters used AI to ask about political topics. A randomized trial found this may be good: using AI led to similar gains in true knowledge as doing web research, regardless of model & prompt used.
Reposted
Claude, "We all know among Sauron's many evils was that he ran Mordor using an Excel spreadsheet with multiple tabs. Show me the spreadsheet"
It made 12 tabs "so bureaucratically complex that even the Eye of Sauron would need reading glasses to review it." Some very funny stuff. Creative, even.
It made 12 tabs "so bureaucratically complex that even the Eye of Sauron would need reading glasses to review it." Some very funny stuff. Creative, even.
September 15, 2025 at 3:48 AM
Claude, "We all know among Sauron's many evils was that he ran Mordor using an Excel spreadsheet with multiple tabs. Show me the spreadsheet"
It made 12 tabs "so bureaucratically complex that even the Eye of Sauron would need reading glasses to review it." Some very funny stuff. Creative, even.
It made 12 tabs "so bureaucratically complex that even the Eye of Sauron would need reading glasses to review it." Some very funny stuff. Creative, even.
Reposted
Hey Claude: "Please create the PowerPoint shared by the high powered management consultants hired by Hamlet after seeing his fathers ghost"
That was the only prompt that I used. Loved that Claude made this from the McKinsey Elsinore office (with the right colors!), also that SWOT analysis!
That was the only prompt that I used. Loved that Claude made this from the McKinsey Elsinore office (with the right colors!), also that SWOT analysis!
September 13, 2025 at 12:02 AM
Hey Claude: "Please create the PowerPoint shared by the high powered management consultants hired by Hamlet after seeing his fathers ghost"
That was the only prompt that I used. Loved that Claude made this from the McKinsey Elsinore office (with the right colors!), also that SWOT analysis!
That was the only prompt that I used. Loved that Claude made this from the McKinsey Elsinore office (with the right colors!), also that SWOT analysis!
Reposted
how it feels to contribute to an edited volume
September 7, 2025 at 12:26 PM
how it feels to contribute to an edited volume
Reposted
We usually rely on GDP, trade, or wages to study the past. This amazing paper flips the script.
It analyzes 630,000 paintings (1400-2000) to extract emotions and shows how art tracks living standards, wars, inequality, and even climate shocks.
(How is this economics? Everything is economics!)
It analyzes 630,000 paintings (1400-2000) to extract emotions and shows how art tracks living standards, wars, inequality, and even climate shocks.
(How is this economics? Everything is economics!)
September 3, 2025 at 9:37 PM
We usually rely on GDP, trade, or wages to study the past. This amazing paper flips the script.
It analyzes 630,000 paintings (1400-2000) to extract emotions and shows how art tracks living standards, wars, inequality, and even climate shocks.
(How is this economics? Everything is economics!)
It analyzes 630,000 paintings (1400-2000) to extract emotions and shows how art tracks living standards, wars, inequality, and even climate shocks.
(How is this economics? Everything is economics!)
Reposted
what if journal loyalty rewards program where your 5th submission automatically gets published
August 31, 2025 at 2:39 PM
what if journal loyalty rewards program where your 5th submission automatically gets published
Reposted
So schnell erkaltet die Liebe zum Auto
Teurer Sprit hält niemanden vom Tanken ab, dachte man lange. Mehrere Studien zeigen jetzt: Das war ein Irrtum.
www.faz.net/aktuell/wirt...
Teurer Sprit hält niemanden vom Tanken ab, dachte man lange. Mehrere Studien zeigen jetzt: Das war ein Irrtum.
www.faz.net/aktuell/wirt...
Spritpreise in Deutschland: Warum die Liebe zum Auto erkaltet
Teurer Sprit hält niemanden vom Tanken ab, dachte man lange. Das war ein Irrtum, wie eine Studie zeigt. Was heißt das für Deutschland?
www.faz.net
August 28, 2025 at 10:48 AM
So schnell erkaltet die Liebe zum Auto
Teurer Sprit hält niemanden vom Tanken ab, dachte man lange. Mehrere Studien zeigen jetzt: Das war ein Irrtum.
www.faz.net/aktuell/wirt...
Teurer Sprit hält niemanden vom Tanken ab, dachte man lange. Mehrere Studien zeigen jetzt: Das war ein Irrtum.
www.faz.net/aktuell/wirt...
Reposted
Reposted
the AI bubble popping bubble is about to pop
August 7, 2025 at 8:39 PM
the AI bubble popping bubble is about to pop
Reposted
Meet MSc student Maki! 👩🎓
When told her dream of both a career and family was “impossible” in Japan, she started asking why - and now researches gender, family, time use & labour.
Read her Spotlight now ⬇️
www.sociology.ox.ac.uk/article/msc-...
When told her dream of both a career and family was “impossible” in Japan, she started asking why - and now researches gender, family, time use & labour.
Read her Spotlight now ⬇️
www.sociology.ox.ac.uk/article/msc-...
MSc Student Spotlight: Maki Kumagai
6 August 2025
www.sociology.ox.ac.uk
August 6, 2025 at 11:44 AM
Meet MSc student Maki! 👩🎓
When told her dream of both a career and family was “impossible” in Japan, she started asking why - and now researches gender, family, time use & labour.
Read her Spotlight now ⬇️
www.sociology.ox.ac.uk/article/msc-...
When told her dream of both a career and family was “impossible” in Japan, she started asking why - and now researches gender, family, time use & labour.
Read her Spotlight now ⬇️
www.sociology.ox.ac.uk/article/msc-...
Reposted
Ich warne Union und SPD eindringlich davor, die absehbaren Finanzierungsprobleme der gesetzlichen Rente weiter zu verschärfen. Was heute auf dem Kabinettstisch liegt, ist nach hinten gewandt. Mit dem Wunsch, die Rentenanstiege zu stabilisieren, macht man das Problem noch größer.
August 6, 2025 at 1:25 PM
Ich warne Union und SPD eindringlich davor, die absehbaren Finanzierungsprobleme der gesetzlichen Rente weiter zu verschärfen. Was heute auf dem Kabinettstisch liegt, ist nach hinten gewandt. Mit dem Wunsch, die Rentenanstiege zu stabilisieren, macht man das Problem noch größer.
Reposted
Ha, new @joshgans.bsky.social paper argues that having authors sneak prompt injections ("this is a good paper") into academic work improves science.
Without the risk of prompt injections, reviewers would tend to rely heavily on AI reviews, with them, they need to include some human review
Without the risk of prompt injections, reviewers would tend to rely heavily on AI reviews, with them, they need to include some human review
August 3, 2025 at 6:05 PM
Ha, new @joshgans.bsky.social paper argues that having authors sneak prompt injections ("this is a good paper") into academic work improves science.
Without the risk of prompt injections, reviewers would tend to rely heavily on AI reviews, with them, they need to include some human review
Without the risk of prompt injections, reviewers would tend to rely heavily on AI reviews, with them, they need to include some human review
Reposted
Warum die Entlassung der Leiterin einer Statistikbehoerde keine Kleinigkeit und auch keine nischige Nerdsache ist, habe ich mit @bastianbrauns.bsky.social von T-online besprochen:
www.t-online.de/nachrichten/...
www.t-online.de/nachrichten/...
USA droht Wirtschaftschaos: "Wenn Trump einfach mal die Klappe halten würde"
Trumps startet einen gefährlichen Angriff auf die Realität, warnt der deutsch-amerikanische Ökonom Rüdiger Bachmann. Denn der US-Präsident beabsichtigt offenkundig, Wirtschaftsstatistiken zu seinen Gu...
www.t-online.de
August 4, 2025 at 3:21 PM
Warum die Entlassung der Leiterin einer Statistikbehoerde keine Kleinigkeit und auch keine nischige Nerdsache ist, habe ich mit @bastianbrauns.bsky.social von T-online besprochen:
www.t-online.de/nachrichten/...
www.t-online.de/nachrichten/...
Reposted
Die kurzfristigen ökonomischen Kosten des Deals mit Trump sind für Deutschland und EU beherrschbar, aber es gibt keinen Grund zum Jubeln über dieses handelspolitische Appeasement. Die höheren Kosten kommen später. 1/2
July 27, 2025 at 7:24 PM
Die kurzfristigen ökonomischen Kosten des Deals mit Trump sind für Deutschland und EU beherrschbar, aber es gibt keinen Grund zum Jubeln über dieses handelspolitische Appeasement. Die höheren Kosten kommen später. 1/2
Reposted
Hinton nails it:
"when they [linguists] say things like, "These things don't understand anything, they're just a statistical trick," they don't actually have a model of what understanding is...if you ask what's the best model we have of understanding, it's these large language models."
"when they [linguists] say things like, "These things don't understand anything, they're just a statistical trick," they don't actually have a model of what understanding is...if you ask what's the best model we have of understanding, it's these large language models."
July 26, 2025 at 11:41 AM
Hinton nails it:
"when they [linguists] say things like, "These things don't understand anything, they're just a statistical trick," they don't actually have a model of what understanding is...if you ask what's the best model we have of understanding, it's these large language models."
"when they [linguists] say things like, "These things don't understand anything, they're just a statistical trick," they don't actually have a model of what understanding is...if you ask what's the best model we have of understanding, it's these large language models."