In 2 digital field experiments, we found that unfollowing hyperpartisan influencers on Twitter: -Reduced partisan animosity by 24%—with effects persisting for 6 months! -Increased satisfaction with Twitter -Led people to share higher-quality news doi.org/10.31234/osf.i…

Here's a summary of our Nature Computational Science paper (led by the amazing Tiancheng Hu & Yara Kyrychenko 🇺🇦) on how LLMs exhibit in-group bias like humans & how to mitigate this by New York University: nyu.edu/about/news-pub…

1/9 🧵 New paper alert (now in Nature Computational Science)! As polarisation continues to shape our world, we asked: Do social and political biases transfer to our AI? I.e. do LLMs show ingroup and outgroup bias? nature.com/articles/s4358…

🚨New Preprint🚨 *Moral Stereotyping in Large Language Models* We show that LLMs are inaccurate in estimating the moral values of diverse cultures: LLMs stereotype cultures in predictable ways. Led by my star Ph.D. student, Aliah Zewail osf.io/preprints/psya…

2024 has been incredible: 15+ publications, and exciting progress in #ResponsibleAI 🎉 Check out our yearly newsletter for all the highlights and what’s next for 2025 social-dynamics.net/newsletter/202…

🟠Delighted to share that my book The Ideological Brain will be coming out in March! 🔵 It’s about what makes brains susceptible to rigid worldviews, how ideologies infiltrate our minds & bodies, and what it means to break free. 📘If you're curious: amazon.com/Ideological-Br…

The first paper from my PhD on #Ukraine wartime digital diplomacy has been published in MWC Journal !💥 journals.sagepub.com/doi/full/10.11… Oxford Internet Institute Green Templeton College University of Oxford

Are LLMs biased when they write about political issues? We just released IssueBench – the largest, most realistic benchmark of its kind – to answer this question more robustly than ever before. Long 🧵with spicy results 👇

🤖🍲 What can LLMs do for sustainable food? 🤖🍲 We collaborated with domain experts (food scientists and chefs) to define a typology of food design and prediction tasks. LLMs can assist in food and menu development, saving food scientists' time and reducing emissions!

1/ So many people have been commenting on the "4% approval" for "dictator" Zelensky. As a sociologist, I can’t resist bringing in some polling data from Ukraine 🧵

I am Ukrainian soldier. I am fighting for my family, my land and my community. I will not put up with bullshit, even if it comes from the American President. I know my values and I am PROTECTING THEM!!!

Ever notice how something that makes your blood boil barely registers with your friend? Our emotional reactions aren't universal at all—they're deeply personal. And AI needs to understand that. Excited to share our new paper: "iNews" 🧵 (1/8) arxiv.org/abs/2503.03335

🕵🏻💬 Introducing Feedback Forensics: a new tool to investigate pairwise preference data. Feedback data is notoriously difficult to interpret and has many known issues – our app aims to help! Try it at app.feedbackforensics.com Three example use-cases 👇🧵

Most emotion detection models treat affect like a universal constant. But emotions are deeply personal. This paper has a dataset that captures affective responses to news posts along with annotator personas. A goldmine for personalization research. arxiv.org/abs/2503.03335

Had a great time on ePODstemology talking with Mark Fabian about my research on socially responsible AI. Tune in! 🎙️🤖

A recent Cambridge University study of more than 65,000 participants from 24 countries measured media consumers' ability to discern mis- and #disinformation to gauge susceptibility, finding that Gen Z, non-male, less educated, and conservative individuals were the groups most

Trending in #SocialPsychology: ooir.org/index.php?fiel… 1) Misinformation susceptibility 2) Beauty & Risk-Taking 3) Authoritarian personality model & punishment preferences(@journal_pops) 4) Self‐Serving Optimism & Well‐Being 5) Illusion of unfairness in random coin flips

Today at #TheWebConf: C3AI – Crafting & Evaluating Constitutions for AI How should we actually write rules for AI? C3AI provides a way to: 1. Design effective constitutions using psychology and public input. 2. Evaluate how well fine-tuned models actually follow the rules.

Ms. Yara Kyrychenko 🇺🇦 of Cambridge University on #EU's AI🤖capabilities: "If Europe had their own #LLM, then other people can start building on it, building on a good🕊️foundation. Otherwise they work with other LLMs, that might be harmful." #RigaStratComDialogue