Sign in

Manuel Tonneau

@manueltonneau.bsky.social
782 followers 572 following 78 posts

Postdoc in AI & Society @hpi.bsky.social, PhD @oii.ox.ac.uk manueltonneau.com

PostsRepliesMedia
Reposted by Manuel Tonneau
Andrew Gordon @andrewgordon.bsky.social · 08/10/2026
This is a super interesting and unique approach to assessing LLM use in online platforms - highly worth a read! 👏
042
Reposted by Manuel Tonneau
Neil Sehgal @nsehgal.bsky.social · 08/10/2026
🚨 New paper 🚨 Are your participants using AI to take your study? We conducted the first study directly observing AI assistance in online research, spanning 700K+ Prolific submissions. We found very rare, concentrated AI use, but gaps in safeguards. arxiv.org/abs/2610.09279
13620
Reposted by Manuel Tonneau
Oxford Internet Institute @oii.ox.ac.uk · 23/09/2026
🎓✨ New Doctor incoming! Congratulations to @manueltonneau.bsky.social on a successful viva this week! His thesis is titled: ‘Examining disparities in online hate speech moderation across the globe’. 1/2
1121
Manuel Tonneau @manueltonneau.bsky.social · 22/09/2026
I have completed my DPhil in Social Data Science @oii.ox.ac.uk 🎓 Deepest thanks to my supervisors Ralph Schroeder and @computermacgyver.bsky.social for their support throughout, and to my examiners @mmosleh.bsky.social and @bjoernross.bsky.social for their careful reading and a lovely discussion!
060
Manuel Tonneau @manueltonneau.bsky.social · 11/09/2026
was pleasantly surprised to have your post linked there as I read the article :)
110
Reposted by Manuel Tonneau
Mohsen Mosleh @mmosleh.bsky.social · 31/08/2026
New preprint: Community Notes are good at correcting posts. But do they change the people behind them? We tracked 19,854 corrected accounts on X — 11.9M posts, 4 weeks before/after each correction. The answer: it depends who you are. @oii.ox.ac.uk 📎 arxiv.org/abs/2608.27526
arxiv.org
Community corrections have divergent downstream effects across corrected accounts
Community-based fact-checking can reduce the spread of annotated misleading posts, but whether it produces lasting behavioral change among corrected authors remains unclear. Here, we conduct a large-s...
2158
Reposted by Manuel Tonneau
The New Yorker @newyorker.com · 08/08/2026
Millions across the world suffer from extreme morning sickness—so severe it leads to dehydration, hospitalization, and even miscarriage. Yet, for decades, the condition was thought to be psychological. www.newyorker.com/news/essay/c...
newyorker.com
Can Extreme Morning Sickness Be Cured?
For decades, the condition was thought to be psychological, and women were shamed for their illness. Now a scientist has found a biological cause.
1234
Reposted by Manuel Tonneau
Jed Brown @jedbrown.org · 17/07/2026
Editor in Chief of Science: AI in scientific publishing: Slower, worse, and more expensive.
doi.org
AI in scientific publishing: Slower, worse, and more expensive
There’s a saying in the management world, popularized by NASA administrator Daniel Goldin in the 1990s, that the goal of technological improvements is to make products faster, better, and cheaper. Alt...
620586
Reposted by Manuel Tonneau
Joachim Baumann @joachimbaumann.bsky.social · 10/07/2026
Our work got featured in Science News! What a great end to an amazing #ICML2026🇰🇷 Peer review is at a crossroads. How we handle the submission and review crisis over the next years will shape the future of scientific publishing. True impact will come from people writing fewer, more meaningful papers!
1172
Reposted by Manuel Tonneau
Weizenbaum-Institut @weizenbauminstitut.bsky.social · 02/07/2026
Weizenbaum researchers were well represented at #FAccT2026 in Montréal, one of the leading conferences on the sociotechnical dimensions of #AI and #computing. Contributions included tutorials, paper presentations, & session chair roles. Congratulations to all involved! 🎉 @facct.bsky.social @acm.org
272
Manuel Tonneau @manueltonneau.bsky.social · 28/06/2026
Sorry things went bad, I'm taking one tomorrow 😅
330
Reposted by Manuel Tonneau
Manoel Horta Ribeiro @manoelhortaribeiro.bsky.social · 18/06/2026
Please join our interview study 🙏
031
Reposted by Manuel Tonneau
Oxford Internet Institute @oii.ox.ac.uk · 09/06/2026
“Social media does not create these tensions but it amplifies and accelerates them and a lot of that amplification is built into how the platform works...” @manueltonneau.bsky.social @oii.ox.ac.uk on the rise of online hate and its spillover into real world violence. www.mirror.co.uk/news/politic...
mirror.co.uk
Social media branded 'tinder box' in aftermath of scandals like Nowak murder
EXCLUSIVE: Online and political experts have warned that social media firms must take greater action to stop the rapid spread of anger on their sites to avoid it spilling onto the streets
042
Reposted by Manuel Tonneau
Stephan Lewandowsky @lewan.uk · 04/06/2026
🧵 Thread: Does the internet threaten democracy? I argue that it does in my latest piece in Science: doi.org/10.1126/scie... Here's the short version. 1/9 1/10
doi.org
The architecture of the internet creates risks for democracy
Will democracy survive the internet? Do we need to choose between Facebook’s surveillance capitalism or democracy? Layered lines of evidence can inform questions like these. When considered together, the evidence gives rise to a concerning picture, as summarized in a recent report for the European Commission that I co-led.
611254
Reposted by Manuel Tonneau
Manoel Horta Ribeiro @manoelhortaribeiro.bsky.social · 03/06/2026
🚨New preprint!🚨 arxiv.org/abs/2606.03095 In a randomized field experiment in a 300-level machine learning course, teaching assistants received AI-assisted feedback drafts after grading student submissions. Did AI help them provide more feedback when feedback was optional? A thread 🧵
194
Reposted by Manuel Tonneau
Joachim Baumann @joachimbaumann.bsky.social · 01/05/2026
Can you boost your AI review scores by asking an LLM to rewrite your paper? Yes! We call it paper laundering Our @icmlconf.bsky.social 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 🧵👇 #ICML2026
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.
44314
Reposted by Manuel Tonneau
Paul Röttger @paul-rottger.bsky.social · 27/04/2026
New paper w/ UK AISI: Millions of people now use AI to help them write and communicate. In three experiments (14k participants, 3m+ human ratings) we show that AI writing assistance systematically distorts writer personas – their perceived beliefs, personality, and identity. 🧵
24413
Reposted by Manuel Tonneau
Lisa Oswald @lfoswaldo.bsky.social · 18/03/2026
I'm hiring a postdoc! @goetheuni.bsky.social Focus: CSS, political behavior, political communication & transforming information environments. 📍 Frankfurt | ⏳ 3 years | 📅 Deadline: 14 April 2026 Full job ad here: www.uni-frankfurt.de/48794987/Zen... (search for “political behavior” to find it)
uni-frankfurt.de
Goethe-Universität — Zentrale Einrichtungen
Die Goethe-Universität ist eine forschungsstarke Hochschule in der europäischen Finanzmetropole Frankfurt. Lebendig, urban und weltoffen besitzt sie als Stiftungsuniversität ein einzigartiges Maß an E...
06351
Reposted by Manuel Tonneau
Workshop on Online Abuse and Harms @woahworkshop.bsky.social · 10/03/2026
WOAH is coming back for its 10th edition at EMNLP 2026 in Budapest! 🎊 For this important anniversary, we invite authors to critically reflect on our achievements as a community and adjust the aim going forward. Stay tuned, more updates coming soon! #EMNLP2026 #WOAH2026 #NLProc
0138
Manuel Tonneau @manueltonneau.bsky.social · 03/03/2026
Thrilled to share that our paper has been accepted to FAccT! See you all in Montréal in June 🇨🇦
050
Reposted by Manuel Tonneau
Francesco Pierri @frapierri.bsky.social · 02/03/2026
Due to the high number of applicants we extended the deadline by one week to **March 8th**. css2.lakecomoschool.org
css2.lakecomoschool.org
Homepage - Computational Social Science
Computational Social Science: Advances, Challenges and Opportunities 2nd Edition June 22-26, 2026 Villa del Grumello, Como, Italy Over the past decade, computational social science (CSS) has risen as ...
011
Reposted by Manuel Tonneau
Philine Widmer @phinifa.bsky.social · 18/02/2026
Can feed algorithms shape what people think about politics? Our paper "The Political Effects of X's Feed Algorithm" is out today in Nature and answers "Yes." www.nature.com/articles/s41...
4273129
Reposted by Manuel Tonneau
David Rand @dgrand.bsky.social · 03/02/2026
🚨New WP "@Grok is this true?" We analyze 1.6M factcheck requests on X (grok & Perplexity) 📌Usage is polarized, Grok users more likely to be Reps 📌BUT Rep posts rated as false more often—even by Grok 📌Bot agreement with factchecks is OK but not great; APIs match fact-checkers osf.io/preprints/ps...
211747
Manuel Tonneau @manueltonneau.bsky.social · 27/01/2026
Thanks also to @johnholbein1.bsky.social whose numerous posts on demographic probing in the social sciences inspired this work and to Matthew Kearney for the useful benchmark dataset.
020
Manuel Tonneau @manueltonneau.bsky.social · 27/01/2026
Lots more info in the paper: arxiv.org/abs/2601.18486 I had a blast working on this with my wonderful coauthors @nsehgal.bsky.social Niyati, Victor, Ana Maria, Lakshmi, Sharath and @valentinhofmann.bsky.social Feedback welcome! @oii.ox.ac.uk
arxiv.org
Demographic Probing of Large Language Models Lacks Construct Validity
Demographic probing is widely used to study how large language models (LLMs) adapt their behavior to signaled demographic attributes. This approach typically uses a single demographic cue in isolation...
120
Manuel Tonneau @manueltonneau.bsky.social · 27/01/2026
Bottom line: LLM demographic probing lacks construct validity: it does not yield a stable characterization of how models condition on demographics. We thus recommend using multiple, ecologically valid cues and controlling for confounders to make defensible claims on demographic effects in LLMs.
110
Manuel Tonneau @manueltonneau.bsky.social · 27/01/2026
Why does this happen? We find that cues differ both in how strongly models associate them with demographic traits and in the non-demographic linguistic features they carry, such as readability or length, and that both independently affect model behavior.
110
Manuel Tonneau @manueltonneau.bsky.social · 27/01/2026
Key result 2: Conclusions on demographic bias depend on how identity is operationalized. Group disparities, estimated as outcome ratios between groups (e.g., Black vs. White), are unstable and vary in magnitude and even direction across cues.
110
Manuel Tonneau @manueltonneau.bsky.social · 27/01/2026
Key result 1: Different cues signalling the same group do not lead to the same model behavior. Cues intended to represent the same demographic group often induce only moderately correlated changes in model behavior.
120
Manuel Tonneau @manueltonneau.bsky.social · 27/01/2026
We study demographic probing in realistic advice-seeking interactions: healthcare, salary, and legal advice, focusing on race and gender in a U.S. context across multiple LLMs. Same prompts. Same tasks. Only the demographic cue signalling group membership changes.
110
Manuel Tonneau @manueltonneau.bsky.social · 27/01/2026
Demographic cues (eg, names, dialect) are widely used to study how LLM behavior may change depending on user demographics. Such cues are often assumed interchangeable. 🚨 We show they are not: different cues yield different model behavior for the same group and different conclusions on LLM bias. 🧵👇
1189
Reposted by Manuel Tonneau
Paul-Antoine @pac.eurosky.social · 23/01/2026
Détection de coordination, détection de #deepfakes, biais et vulnérabilités des algorithmes de recommandation... @viginum.bsky.social et #INRIA lancent un prix scientifique de lutte contre les manipulations de l'information. 👉 pvi-lmi.sciencescall.org Deadline le 14/02. #disinfo #FIMI
pvi-lmi.sciencescall.org
Prix Viginum - Inria « Lutte contre les manipulations de l'information » - Sciencesconf.org
Le Prix VIGINUM-Inria
02722
Manuel Tonneau @manueltonneau.bsky.social · 21/01/2026
Kudos to my wonderful co-authors Do Lee doqlee.github.io , Boris Sobol il.linkedin.com/in/boris-sobol , @nirg.bsky.social and Sam Fraiberger samuelfraiberger.com. @oii.ox.ac.uk @nyupress.bsky.social 11/fin
020
Manuel Tonneau @manueltonneau.bsky.social · 21/01/2026
Yet platform data-access policies increasingly block this potential. Whether platforms or regulators will enable change in the coming years is a core policy question. 10/N
110
Manuel Tonneau @manueltonneau.bsky.social · 21/01/2026
There is clear public value here, potentially extending to other countries, especially where official statistical systems are under-developed. 9/N
110
Manuel Tonneau @manueltonneau.bsky.social · 21/01/2026
Why this matters? Beyond forecasting, this approach can provide early warnings, surface local labor market stress hidden by national averages, and help flag measurement issues in real time. 8/N
120
Manuel Tonneau @manueltonneau.bsky.social · 21/01/2026
Key finding 3: This also works at the state and city (!) level, including "holdout cities" where official UI numbers are sparse or irregularly updated. As expected, accuracy scales with platform penetration and unemployment shocks. 7/N
120
Manuel Tonneau @manueltonneau.bsky.social · 21/01/2026
Key finding 2: Our approach consistently outperforms industry consensus forecasts and can improve predictions of US UI claims up to two weeks ahead of official releases. That’s two weeks of additional lead time for policymakers. 6/N
110
Manuel Tonneau @manueltonneau.bsky.social · 21/01/2026
Key finding 1: Capturing linguistic diversity matters. Training LLMs with active learning lets us detect many more ways people talk about job loss, producing a far more representative sample of unemployed users than existing approaches. 5/N
110
Manuel Tonneau @manueltonneau.bsky.social · 21/01/2026
We combine JoblessBERT (an encoder LLM developed in previous work aclanthology.org/2022.acl-lon... which detects ~3× more employment-related content without sacrificing precision) with post-stratification using inferred demographics to correct for platform bias. 4/N
aclanthology.org
Multilingual Detection of Personal Employment Status on Twitter
Manuel Tonneau, Dhaval Adjodah, Joao Palotti, Nir Grinberg, Samuel Fraiberger. Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2022.
120
Manuel Tonneau @manueltonneau.bsky.social · 21/01/2026
So we ask a hard question economic actors and policymakers rightly worry about: Can skewed social media data be turned into trustworthy indicators of unemployment? Can we produce robust predictions across geography ✅, time ✅, demography ✅, and forecasting horizon ✅ ? 3/N
120
Manuel Tonneau @manueltonneau.bsky.social · 21/01/2026
Why this matters: In March 2020, weekly unemployment insurance claims jumped from 278K to nearly 6 million in two weeks. As official data lagged, policymakers were flying blind about where the shock was hitting and who was being affected. 2/N
110
Manuel Tonneau @manueltonneau.bsky.social · 21/01/2026
🚨 New paper out in @pnasnexus.org We show how skewed social media data can still be used to reliably estimate unemployment, not just nationally but down to the city level. 📈 doi.org/10.1093/pnas... 1/N
doi.org
Can social media reliably estimate unemployment?
Abstract. Digital trace data hold tremendous potential for measuring policy-relevant outcomes in real-time, yet its reliability is often questioned. Here,
173
Reposted by Manuel Tonneau
PNAS Nexus @pnasnexus.org · 02/01/2026
A transformer encoder-based classifier called JoblessBERT can identify posts about unemployment on social media, allowing researchers to predict US unemployment claims, up to two weeks in advance, at the national, state, and city levels. In PNAS Nexus: ow.ly/Zvi850XRa8I
Illustration of the official unemployment rate published in newspapers. Stock photo.
021
Reposted by Manuel Tonneau
Oxford Internet Institute @oii.ox.ac.uk · 01/10/2025
ICYMI: Listen to @manueltonneau.bsky.social @oii.ox.ac.uk's interview with the SOEP podcast talking about his new research into hate speech, online platforms and disparities in content moderation across different European countries. Available here: bit.ly/4ntsiRU
bit.ly
Home - Somewhere On Earth Productions
SOMEWHERE ON EARTH PRODUCTIONS: We are here to connect technology and business to people and new possibilities.
011
Reposted by Manuel Tonneau
Andreu Casas @andreucasas.bsky.social · 29/09/2025
🚨Hiring a fully funded (3.5 years) PhD for the @ldnsocmedobs.bsky.social to research social media and politics. Candidates should have quantitative/computational skills and/or be interested in content curation/moderation. UK home candidates only unfortunately. www.royalholloway.ac.uk/media/hquftp...
royalholloway.ac.uk
2514
Reposted by Manuel Tonneau
Tanise Ceron @taniseceron.bsky.social · 29/09/2025
📣 New Preprint! Have you ever wondered what the political content in LLM's training data is? What are the political opinions expressed? What is the proportion of left- vs right-leaning documents in the pre- and post-training data? Do they correlate with the political biases reflected in models?
24613
Reposted by Manuel Tonneau
Manoel Horta Ribeiro @manoelhortaribeiro.bsky.social · 16/09/2025
Social media feeds today are optimized for engagement, often leading to misalignment between users' intentions and technology use. In a new paper, we introduce Bonsai, a tool to create feeds based on stated preferences, rather than predicted engagement. arxiv.org/abs/2509.10776
516146
Reposted by Manuel Tonneau
Joachim Baumann @joachimbaumann.bsky.social · 12/09/2025
🚨 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.
6303106
Reposted by Manuel Tonneau
Chris Bail @chrisbail.bsky.social · 29/08/2025
1/ 🚨 Big news 🚨 today we’re launching Tech for Open Minds (TOM) at @DukeU— a global program exploring how technology shapes open-mindedness, humility & polarization 🌍🧠 🔗https://sicss.io/stories/2025-08-18
1102