Maximilian Maurer
LightNews LightNews
mmmaurer.bsky.social
Maximilian Maurer
@mmmaurer.bsky.social
200 followers 350 following 13 posts
PhD student at @gesis.org & @hhu.de, computational linguist, researching linguistic factors in (annotation) disagreement and language model behavior.
Posts Replies Media Videos
Reposted by Maximilian Maurer
anids.bsky.social
Every day a new reason to hate Outlook
November 12, 2025 at 10:22 PM
Reposted by Maximilian Maurer
mtutek.bsky.social
*Urgently* looking for emergency reviewers for the ARR October Interpretability track 🙏🙏

ReSkies much appreciated
November 11, 2025 at 10:29 AM
mmmaurer.bsky.social
happening in 10min, poster board 40!
November 5, 2025 at 2:50 AM
mmmaurer.bsky.social
Does AI argue differently than humans? 🤖
The results of our #EMNLP2025 paper suggest so!

@esradonmez.bsky.social and I will present this work in Poster Session 1, 11:00-12:30 on 5th Nov, Hall C3

🧵👇

aclanthology.org/2025.emnlp-m...
Kernel density estimation plots showing differences between human comments and LLMs (GPT, Mistral, and Llama), for two argument quality features (reference and impact) and two linguistic features (TTR and average anticipation). The plots show more variability of humans and more narrow distributions for LLMs, respectively.
November 3, 2025 at 11:55 PM
Reposted by Maximilian Maurer
gesistraining.bsky.social
*Preprocessing and Analyzing Web Tracking Data* by Frank Mangold, Helena Rauxloh & @sebstier.bsky.social

🗓️ 04-05 May 2026
🏢 Hybrid: Cologne | Online
🌐 t1p.de/WebTrackingD...
November 3, 2025 at 1:28 PM
Reposted by Maximilian Maurer
mariaa.bsky.social
Here’s a custom feed for #EMNLP2025. Click the pin to save it to your home screen!
November 2, 2025 at 3:15 PM
Reposted by Maximilian Maurer
python.org
TLDR; The PSF has made the decision to put our community and our shared diversity, equity, and inclusion values ahead of seeking $1.5M in new revenue. Please read and share. pyfound.blogspot.com/2025/10/NSF-...
🧵
The official home of the Python Programming Language
www.python.org
October 27, 2025 at 2:47 PM
Reposted by Maximilian Maurer
lottepuettmann.bsky.social
„manche stehen schon in flammen
andere riechen nicht einmal den rauch“
(Max Czollek)
February 1, 2025 at 9:35 AM
Reposted by Maximilian Maurer
maxczollek.bsky.social
Ach und er sagt auch noch, dass Shitstorm ein modernes Wort für Pogrome ist. Genau! Und Jana aus Kassel ist Sophie Scholl wegen Kritik. Es ist EXAKT DASSELBE DENKEN wie bei den Schwurblern, die sich zur Pandemie einen Judenstern angeheftet haben. Mach mal Fenster auf Kipp, Richard.
October 17, 2025 at 6:41 AM
Reposted by Maximilian Maurer
maxczollek.bsky.social
Darf ich vorstellen: die vegane Schinkenwurst der Faschismusanalysen. Wenn die EU das bei Lebensmitteln verbietet, sollte sie bei Richard David Prechts Denkimitation weitermachen. Das ist nämlich weit überm Verfallsdatum.
Schlagzeile: Richard David Precht. "Jemanden zu canceln, ist für mich Faschismus"
October 17, 2025 at 6:36 AM
Reposted by Maximilian Maurer
christinaviehmann.bsky.social
​The #GESISMethodsHub is live! 🎉
​So thrilled to launch this project, built with so much heart. It's set to be your community portal for computational tools and coding environments.
​Explore now! We'll open for external submissions by year-end.
gesis.org GESIS - Leibniz-Institut für Sozialwissenschaften @gesis.org · Oct 15
#DataMethods #OpenScience

#GESISMethodsHub: An open portal for computational social science research—find tools, tutorials, and code, directly usable in your browser. Methods Hub makes it easy to learn, apply, and reproduce computational methods in the Social Sciences.

methodshub.gesis.org/
October 15, 2025 at 3:00 PM
Reposted by Maximilian Maurer
gesis.org
#DataMethods #OpenScience

#GESISMethodsHub: An open portal for computational social science research—find tools, tutorials, and code, directly usable in your browser. Methods Hub makes it easy to learn, apply, and reproduce computational methods in the Social Sciences.

methodshub.gesis.org/
October 15, 2025 at 8:00 AM
Reposted by Maximilian Maurer
sebstier.bsky.social
📢 We're hiring a Doctoral Student in Research Software Engineering for Digital Behavioral Data!
Work at the intersection of computer science and social science, develop innovative tools, and pursue your doctoral degree with us.

👉 Apply here: www.gesis.org/en/institute...
Details
GESIS Leibniz Institut für Sozialwissenschaften
www.gesis.org
October 2, 2025 at 7:48 AM
Reposted by Maximilian Maurer
kwelle.bsky.social
Social-Media-Daten zwischen Forschung und Infrastrukturen - nachhaltige Archivierung, Erschließung und Bereitstellung: An der @dnb-aktuelles.bsky.social finden vom 17.-19.03.2026 die Social Media Access Days statt. Wir freuen uns über Einreichungen bis zum 31.10.2025. www.dnb.de/DE/Professio...
Call for Submissions: Social Media Access Days
Call for Submissions: Social Media Access Days
www.dnb.de
October 1, 2025 at 6:29 AM
Reposted by Maximilian Maurer
joachimbaumann.bsky.social
🚨 New paper alert 🚨 Using LLMs as data annotators, you can produce any scientific result you want. We call this **LLM Hacking**.

Paper: arxiv.org/pdf/2509.08825
We present our new preprint titled "Large Language Model Hacking: Quantifying the Hidden Risks of Using LLMs for Text Annotation". We quantify LLM hacking risk through systematic replication of 37 diverse computational social science annotation tasks. For these tasks, we use a combined set of 2,361 realistic hypotheses that researchers might test using these annotations. Then, we collect 13 million LLM annotations across plausible LLM configurations. These annotations feed into 1.4 million regressions testing the hypotheses. For a hypothesis with no true effect (ground truth $p > 0.05$), different LLM configurations yield conflicting conclusions. Checkmarks indicate correct statistical conclusions matching ground truth; crosses indicate LLM hacking -- incorrect conclusions due to annotation errors. Across all experiments, LLM hacking occurs in 31-50\% of cases even with highly capable models. Since minor configuration changes can flip scientific conclusions, from correct to incorrect, LLM hacking can be exploited to present anything as statistically significant.
September 12, 2025 at 10:33 AM
Reposted by Maximilian Maurer
nlp-amelie.bsky.social
Crowdsourcing datasets is very common, but how much of those results come from LLMs these days? How can we prevent it? Much of it is unclear. This is why we’re conducing a survey (~10 min) to gather community experiences, challenges, and solutions.
Share your thoughts 👉https://tinyurl.com/39ab55wf
Qualtrics Survey | Qualtrics Experience Management
The most powerful, simple and trusted way to gather experience data. Start your journey to experience management and try a free account today.
ugent.qualtrics.com
September 5, 2025 at 11:32 AM
Reposted by Maximilian Maurer
cerveza-inglesa.bsky.social
🌈 Help with my LGBTQIA+ research! 🌈

I'm building a dataset to fight harmful LGBTQIA+ stereotypes in AI. Take my quick survey:
👉 IT: lnkd.in/dfPuyT6j
👉 NL: lnkd.in/d-3Di7WY
👉 FA: lnkd.in/dfvWzWCu
Everybody is welcome!

📤 Share if you can!
July 21, 2025 at 10:37 AM
Reposted by Maximilian Maurer
div-ling.org
One highlight of LILG 2025: the official launch of the new open-access journal Gender Linguistics!

Published by our association, the journal will be a space for research at all linguistic intersections of language and gender.

Learn more: gender-linguistics.org

#linguistics #journal #launch
on the left, the cover of the journal gender linguistics, its logo consists of a conjoined uppercase G and L. on the right, the following text: a community-owned journal for gender linguistic research, peer-reviewed, open access, no publication costs, multilingual publishing; below the text, there is the link to the journal website and a qr code.
August 14, 2025 at 3:26 PM
Reposted by Maximilian Maurer
nsaphra.bsky.social
The transformation of the irreplaceable treasure “performative” into a mere synonym for “phony” is a lexical tragedy to be mourned by future generations
August 11, 2025 at 12:57 AM
Reposted by Maximilian Maurer
markriedl.bsky.social
The complete lack of regard for the Y-axis hurts me physically #GPT5
aparker.io austin @aparker.io · Aug 7
data visualization is my passion
clip from a openai presentation showing swe bench performance; the openai o3 and gpt-4o models are at the same visual level but the labels are wildly different
August 7, 2025 at 5:32 PM
Reposted by Maximilian Maurer
gabriellalapesa.bsky.social
Identity-Aware AI workhop at #ECAI2025, in beautiful Bologna! Submit by Aug 22: identity-aware-ai.github.io Organized by:
@pranav-nlp.bsky.social Soda Marem Lo, Neele Falk, @gingerinai.bsky.social @davidjurgens.bsky.social @a-lauscher.bsky.social and myself!
ds-hamburg.bsky.social Data Science Hamburg Group @ds-hamburg.bsky.social · Jul 29
Wondering what makes each of us unique and how AI should handle human diversity? 🤔

We're organizing Identity-Aware AI workshop at #ECAI2025 Bologna on Oct 25.

Deadline: Aug 22
Website: identity-aware-ai.github.io
Identity-Aware AI Workshop announcement. Co-located with ECAI 2025 in Bologna on October 25, with submission deadline August 22. Topics include: Methods for effective, fair, and inclusive AI; Critiques of AI on the exclusion of identities; Methods for detecting and controlling bias; Perspectivist approaches to AI. Submission types: Long papers (8 pages), Short papers (4 pages), Extended abstracts, Mixed-media submissions (videos, blogs, codebase, artworks). For details, visit: identity-aware-ai.github.io
August 7, 2025 at 3:29 PM
Reposted by Maximilian Maurer
timkellogg.me
word is GPT-5 will have a 13.5% increase in version number over GPT-4.5
August 6, 2025 at 4:18 PM
Reposted by Maximilian Maurer
emollick.bsky.social
We have been carefully testing lots of received prompting wisdom & for recent AI models:
🚫Threats, saying please, being insulting, & promising tips do not change average performance on challenging tasks
⛓️Chain-of-thought no longer helps even non-reasoner performance much

Don't make it complicated.
August 2, 2025 at 7:42 PM
Reposted by Maximilian Maurer
sxu3.bsky.social
👏 Upcoming workshop: Perspectivist Approaches to NLP @EMNLP 2025
Focusing on non-aggregated datasets and multi-perspective modeling, with sessions on labeling, modeling, evaluation, and applications.
nlperspectives.di.unito.it
NLPerspectives – Perspectivist Approaches to NLP
nlperspectives.di.unito.it
July 31, 2025 at 2:50 PM
Reposted by Maximilian Maurer
juand-r.bsky.social
Why all the interest in AI for automating idea generation in science? Isn't this the fun part? We should be automating the tedious parts.
August 1, 2025 at 6:26 PM