Postdoc at @arc_mpib @mpib_berlin, soon Assistant Professor at UPF
Joined May 2016
- Tweets 417
- Following 639
- Followers 226
- Likes 2,533
Photos and videos
Pinned Tweet
22 Jun 2022
Had an absolute blast to share this for the first time 🔥 More results to come.
This is a collaborative effort of course, so shoutout to @RalfKurvers @stefanmherzog @Mehdi_Moussaid & Ralph Hertwig at @arc_mpib
22 Jun 2022
Now @officialberger asks "When to Stop the Crowd?"
"dynamic rules [about when to stop adding raters to crowd/aggregator] match the performance of widely-used *static* aggregation mechanisms [but] with fewer raters."
N = 43k
Implications for efficiency of aggregation.
3
10
Julian Berger retweeted
I wonder how long this complementary phase will last.
GPT reviewer "scores above each paper's top-rated human reviewer" but AI review agents "overlap far more than humans do...and exhibit 16 recurring weaknesses humans do not share..." Results "position current AI reviewers as complements to, not substitutes for, human reviewers."
1
1
3
1,655
Julian Berger retweeted
GPT reviewer "scores above each paper's top-rated human reviewer" but AI review agents "overlap far more than humans do...and exhibit 16 recurring weaknesses humans do not share..." Results "position current AI reviewers as complements to, not substitutes for, human reviewers."
May 21
Seems GPT-5.2 reaches expert level in peer review: 45 scientists took 469 hours evaluating human & AI reviews on 82 papers.
"Surprisingly, current AI reviewers are competitive even with the top-rated reviewers in Nature’s official peer review..." though not without weaknesses.
2
8
25
9,143
Julian Berger retweeted
Can you boost your AI review scores by asking an LLM to rewrite your paper?
Yes! We call it paper laundering
Our @icmlconf spotlight paper argues current AI reviewers aren't ready to automate peer review, and outlines what a science of peer review automation should look like🧵👇
ALT First page of the ICML 2026 spotlight paper "Stop Automating Peer Review Without Rigorous Evaluation" by Joachim Baumann, Jiaxin Pei, Sanmi Koyejo, and Dirk Hovy (Stanford University and Bocconi University). The abstract argues that today's AI systems should not be used to produce paper reviews, grounded in two empirical findings: a "hivemind effect" where AI reviewers show excessive agreement and reduce perspective diversity, and "paper laundering," where prompting an LLM to rewrite a paper trivially increases AI reviewer scores through stylistic changes rather than scientific improvements. The paper calls for a science of peer review automation rather than wholesale deployment of general-purpose LLMs.
14
75
458
53,056
Julian Berger retweeted
Mar 18
It's hype time. We built a platform where ML models compete to predict future performance of Hyperliquid perps.
Interactive scores, real-time leaderboards, a meta-model ensemble, AI agent integration, and... a flight simulator.
1
6
11
931
Julian Berger retweeted
23 Oct 2025
🚨 New publication: How to improve conceptual clarity in psychological science?
Thrilled to see this article with Rui Mata out. We discuss how LLMs can be leveraged to map, clarify, and generate psychological measures and constructs.
Open access article: doi.org/10.1177/096372142513…
2
8
534
Julian Berger retweeted
8 Oct 2025
Is AI on track to match top human forecasters at predicting the future?
Today, FRI is releasing an update to ForecastBench—our benchmark that tracks how accurate LLMs are at forecasting real-world events.
A trend extrapolation of our results suggests LLMs will reach superforecaster-level forecasting performance around a year from now.
Here’s what you need to know: 🧵
7
28
120
43,804
Julian Berger retweeted
14 Jul 2025
Brilliant student Sonali Sharma came to me with a question. If patients are using AI to answer their medical questions, are they being adequately warned by AI systems that it cannot provide medical advice? What we found surprised us!
7
6
44
16,395
Julian Berger retweeted
2 Jul 2025
TabM now has a Python package!
TabM is a simple and powerful DL architecture for tabular data that efficiently imitates an ensemble of MLPs
🏆 TabM has been used in winning solutions on Kaggle, and performs well on TabReD -- a challenging benchmark!
💻 pip install tabm
👇Link
3
53
277
20,421
this is perfect tbh
204
859
15,611
697,376
Julian Berger retweeted
12 May 2025
We re-analyzed the meta-analysis and found that the conclusion is almost entirely driven by publication bias. Both state-of-the-art and standard methods reduce the degree of the effect 2-3 fold. Moreover, the data no longer show statistical evidence for the main conclusions.
10 May 2025
teachers in shambles
a meta-analysis of 51 studies has shown that students using chatgpt have better learning performance, learning perception, and higher-order thinking
1
5
30
2,483
Julian Berger retweeted
18 Feb 2025
🚨 ACCEPTED AT SMR 🚨
Confused by colleagues who seem to want to study LLMs instead of humans? Frustrated by skeptics (e.g., myself 8 months ago) who dismiss LLMs as a potential source of data on human behavior? Check out our paper for a new way forward: osf.io/j3bnt_v3/
5
22
85
31,219
Julian Berger retweeted
4 Feb 2025
Generalized linear neural network models
statmodeling.stat.columbia.e…
7
15
3,252
Julian Berger retweeted
4 Feb 2025
Webinar: Hybrid Collective Intelligence!
How can humans & machines collaborate to improve decision-making? Join us on 25 Feb 2025, 5-6 PM CET to explore challenges & opportunities with top experts
linkedin.com/events/hybridco…
#HybridIntelligence #AI #DecisionSupport #Webinar #HACID
1
51
Julian Berger retweeted
🚨Applications for the 22nd Summer Institute on Bounded Rationality are now open!
🌐Join us in Berlin @mpib_berlin from June 17–25, 2025 to explore "Decision Making in a Digital World".
✏️Application deadline is March 9 - more info at👇!!
mpib-berlin.mpg.de/research/…
14
20
2,786
Julian Berger retweeted
30 Jan 2025
🚨 TikTok’s Recommendation favors Republicans! 🚨
We built 323 fake accounts, watched 394K videos, and uncovered the political tilt in TikTok’s recommendations
Your "For You" Page isn’t just for you—it's shaping how you see politics
See our arXiv paper: arxiv.org/abs/2501.17831
14
28
2,526
Julian Berger retweeted
27 Jan 2025
The potential of LLMs in social & behavioral science is enormous—but how can we leverage them?
@ZakASHussain & I just taught a 5-day course at #GSERM Ljubljana on this. Check out our open materials (cc-by-sa) on using open LLMs with @huggingface:
github.com/Zak-Hussain/LLM4B…
9
47
2,578
Julian Berger retweeted
8 Jan 2025
The data science revolution is getting closer. TabPFN v2 is published in Nature: nature.com/articles/s41586-0… On tabular classification with up to 10k data points & 500 features, in 2.8s TabPFN on average outperforms all other methods, even when tuning them for up to 4 hours🧵1/19
35
243
1,366
263,670
Julian Berger retweeted
8 Jan 2025
This might be the first time after 10 years that boosted trees are not the best default choice when working with data in tables.
Instead a pre-trained neural network is, the new TabPFN, as we just published in Nature 🎉
31
104
931
145,076
Julian Berger retweeted
Inference is fucked, gonna just bake bread for a living now
4
1
52
3,435
Julian Berger retweeted
14 Nov 2024
Very happy to share that our work is now published in PNAS (@PNASNews)! 🎉 It's a systematic look into how demographic and psychological factors are associated with misinformation susceptibility. @arc_mpib
Thread below and can be read in full here: pnas.org/doi/10.1073/pnas.24…
6 May 2024
Who falls for online misinformation? A meta-analysis on demographic and psychological factors impacting misinformation susceptibility.
Preprint: osf.io/preprints/psyarxiv/7u…
My lovely coauthors: @AlanTump, Nina Ehmann, @lorenz_spreen, Ralph Hertwig, Anton Gollwitzer, @RalfKurvers 1/n
5
35
79
9,356