Taha Yasseri
@tahayasseri.bsky.social
Workday Full Prof & Chair of Technology & Society at Trinity College Dublin & Technological University Dublin. Director of the Centre for Sociology of Humans and Machines. https://tahayasseri.com
Our new @poeticsjournal.bsky.social paper disentangles the role of within-individual and compositional changes in shifting political meaning-making on social media following focus events like terrorist attacks: www.sciencedirect.com/science/arti... Together with @mhbodell.bsky.social
The CRT Team are delighted to congratulate Saeedeh Mohammadi on successfully passing her PhD Viva! Her PhD thesis is titled "Human and Machine Collective Intelligence for Content Moderation." Well done Saeedeh! 🎉@researchireland.ie @tahayasseri.bsky.social
Chapter 20: @diyiyang.bsky.social & @calebziems.com on Learning with Weak Supervision for CSS Clear & timely introduction to methods that help researchers overcome 1 of biggest bottlenecks in CSS: lack of labelled data. From distant supervision to data augmentation, semi-supervised learning & LLMs!
Taha Yasseri@tahayasseri.bsky.social · last mo.Chapter 19: @lajello.bsky.social on Learning the Social Pragmatics of Language. A thought-provoking chapter arguing that CSS should move beyond syntax and semantics to study intentions, relationships, and social action expressed through language. A compelling vision for next generation of NLP & CSS.
How you treat an AI agent determines the results you'll get, says Professor Taha Yasseri. Phil Wainewright was at a recent Workday event in London to find out why, and here he shares more findings from the Joint Centre for Sociology of Humans and Machines: bit.ly/4rjH9iU
Trust in science seems compatible with believing almost anything, so long as it is expressed in plausibly scientific terms, write Michele d'Errico and Taha Yasseri. theconversation.com/why-do-some-...
Why do some people who trust in science also believe in scientific conspiracy theories?
Trust in science seems compatible with believing almost anything, so long as it is expressed in plausibly scientific terms.
theconversation.com
Chapter 19: @lajello.bsky.social on Learning the Social Pragmatics of Language. A thought-provoking chapter arguing that CSS should move beyond syntax and semantics to study intentions, relationships, and social action expressed through language. A compelling vision for next generation of NLP & CSS.
Taha Yasseri@tahayasseri.bsky.social · 2mo agoChapter 18: @janlo.de on Exploring Theory with Agent-Based Modeling and Simulation. Showing how agent-based models can move CSS beyond prediction toward explanation, revealing how simple behavioural rules can generate complex social phenomena such as segregation, polarization, and filter bubbles.
Chapter 18: @janlo.de on Exploring Theory with Agent-Based Modeling and Simulation. Showing how agent-based models can move CSS beyond prediction toward explanation, revealing how simple behavioural rules can generate complex social phenomena such as segregation, polarization, and filter bubbles.
Taha Yasseri@tahayasseri.bsky.social · 2mo agoChapter 17: Andreas Flache @mrnkzr.bsky.social & Michael Mäs on Agent-Based Models & Networks. A thoughtful chapter arguing that computational social science needs both large-scale empirical & and mechanism-based simulations to explain how individual interactions produce collective phenomena.
"Researchers can't access social media data anymore." But one of the richest open datasets available today, Community Notes, remains surprisingly understudied. We analyse its first four years and release a research-ready dataset, code, interaction networks, and a comprehensive literature review!
Chapter 17: Andreas Flache @mrnkzr.bsky.social & Michael Mäs on Agent-Based Models & Networks. A thoughtful chapter arguing that computational social science needs both large-scale empirical & and mechanism-based simulations to explain how individual interactions produce collective phenomena.
Taha Yasseri@tahayasseri.bsky.social · 2mo agoChapter 16: Stephanie Zonszein, PM Aronow, & @cdsamii.bsky.social on estimating causal effects in the presence of network spillovers. A clear introduction to one of the key methodological challenges in CSS: how to identify causal effects when one person's treatment changes someone else's outcome.
Today marks the beginning of Ireland's EU Presidency! Watch our video series 'Ireland as EU President' at the link below & see what Ireland can bring to the table over the next six months. www.tcd.ie/ssp/research... @tahayasseri.bsky.social @farbod-a.bsky.social @tcdeconomics.bsky.social
Chapter 16: Stephanie Zonszein, PM Aronow, & @cdsamii.bsky.social on estimating causal effects in the presence of network spillovers. A clear introduction to one of the key methodological challenges in CSS: how to identify causal effects when one person's treatment changes someone else's outcome.
Taha Yasseri@tahayasseri.bsky.social · 3mo agoChapter 15: @kristinalerman.bsky.social on the Strong Friendship Paradox. Showing how the structure of social networks systematically biases our perceptions of reality. When our friends are not representative of the population, rare behaviours and opinions can appear surprisingly common.
Can Ireland keep AI under control?🧐 In our final video of the 'Ireland as EU President' series, Professor @tahayasseri.bsky.social explains why Ireland is the perfect candidate for steering the future of AI regulation in Europe 👇 @tcdsociology.bsky.social @tcddublin.bsky.social
💥New | Gendered AI design reflects and reinforces society’s biases ✍️ @tahayasseri.bsky.social & @cuihaosabrina.bsky.social #Sociology #AI #GenderedDesign
Gendered AI design reflects and reinforces society’s biases - LSE Impact
When AI tools are designed with a gender, they don't just encode gendered values they also encourage users to engage with them in biased ways.
blogs.lse.ac.uk
Many are familiar with Friendship Paradox (“your friends are more popular than you, on average “). But did you know FP holds when you swap mean for the median? Yes, MOST friends are more popular than you. This opens networks up to some mind blowing consequences
Taha Yasseri@tahayasseri.bsky.social · 3mo agoChapter 15: @kristinalerman.bsky.social on the Strong Friendship Paradox. Showing how the structure of social networks systematically biases our perceptions of reality. When our friends are not representative of the population, rare behaviours and opinions can appear surprisingly common.
Chapter 15: @kristinalerman.bsky.social on the Strong Friendship Paradox. Showing how the structure of social networks systematically biases our perceptions of reality. When our friends are not representative of the population, rare behaviours and opinions can appear surprisingly common.
Taha Yasseri@tahayasseri.bsky.social · 3mo agoChapter 14: @bolozna.bsky.social on multilayer social networks. A great introduction to one of the most important recent developments in network science: moving beyond single-layer graphs to represent the multiple social contexts and relationships that shape human behavior. Social life is layered!
Chapter 14: @bolozna.bsky.social on multilayer social networks. A great introduction to one of the most important recent developments in network science: moving beyond single-layer graphs to represent the multiple social contexts and relationships that shape human behavior. Social life is layered!
Taha Yasseri@tahayasseri.bsky.social · 3mo agoChapter 13: @jsaramak.bsky.social & @pholme.bsky.social on temporal networks of social interactions. A fascinating chapter showing why static network snapshots often miss the most important thing: the timing and order of interactions. Social networks are dynamic processes, not just fixed structures.
Chapter 13: @jsaramak.bsky.social & @pholme.bsky.social on temporal networks of social interactions. A fascinating chapter showing why static network snapshots often miss the most important thing: the timing and order of interactions. Social networks are dynamic processes, not just fixed structures.
Taha Yasseri@tahayasseri.bsky.social · 4mo agoChapter 12: @fedebotta.bsky.social from @exetercompsci.bsky.social on online images and computational social science. A fascinating chapter on how images shared online can help us study politics, misinformation, culture, & collective behaviour and why visual data deserves a much bigger place in CSS.
Who'd have thought Grokipedia had a rightward political bias — especially on pages about religion, history, and literature and art? 🤔 And as a bonus, it is harder to read! The analyzed Grokipedia articles in this study were more wordy, complex, and contained fewer references per word.
Selective divergence between Grokipedia and Wikipedia articles | PNAS
The launch of Grokipedia, an AI-generated encyclopedia developed by xAI, was presented as a response to perceived ideological and structural biases...
pnas.org
Selective divergence between Grokipedia and Wikipedia articles www.pnas.org/doi/abs/10.1...
PNAS
Proceedings of the National Academy of Sciences (PNAS), a peer reviewed journal of the National Academy of Sciences (NAS) - an authoritative source of high-impact, original research that broadly spans...
pnas.org
Professor @tahayasseri.bsky.social and PhD candidate Saeedeh Mohammadi have published new research on Grokipedia and its structural biases. 🗣️ ✍️Read the article: www.pnas.org/doi/10.1073/... @tcdschoolssp.bsky.social @soham-centre.bsky.social
A large-scale analysis of Grokipedia, the world’s first AI-written encyclopedia, found that Grokipedia selectively draws on more right-leaning news sources. The study was conducted by Saeedeh Mohammadi and @tahayasseri.bsky.social at @soham-centre.bsky.social. See link to full press release ⬇️
Kicking off day 2 of Machine+Behavior: @tahayasseri.bsky.social moderating a discussion with speakers Kinga Makovi and Eric Schulz.
Paper out in @pnas.org We compared ~18,000 matched articles between Wikipedia and Grokipedia, xAI’s AI-generated encyclopedia. While many pages closely mirrored Wikipedia, a substantial subset diverged markedly in content, sourcing, and political orientation, especially in religion and history.
Paper out in @pnas.org We compared ~18,000 matched articles between Wikipedia and Grokipedia, xAI’s AI-generated encyclopedia. While many pages closely mirrored Wikipedia, a substantial subset diverged markedly in content, sourcing, and political orientation, especially in religion and history.
Chapter 12: @fedebotta.bsky.social from @exetercompsci.bsky.social on online images and computational social science. A fascinating chapter on how images shared online can help us study politics, misinformation, culture, & collective behaviour and why visual data deserves a much bigger place in CSS.
Taha Yasseri@tahayasseri.bsky.social · 4mo agoChapter 11: Luis-Daniel Ibáñez, Johanna Walker & @elenasimperl.bsky.social on open data in computational social science. Why open data is not just a technical issue, but also a question of sustainability, impact, politics, and bias. Useful data is more than available data. @aiatkings.bsky.social
Chapter 11: Luis-Daniel Ibáñez, Johanna Walker & @elenasimperl.bsky.social on open data in computational social science. Why open data is not just a technical issue, but also a question of sustainability, impact, politics, and bias. Useful data is more than available data. @aiatkings.bsky.social
Taha Yasseri@tahayasseri.bsky.social · 4mo agoChapter 10: @feloe.bsky.social & @vanatteveldt.com on social media data donation and digital tracking. A very useful chapter on how digital traces can be integrated into social science research, not as a replacement for surveys and other methods, but as a powerful complement.
Chapter 10: @feloe.bsky.social & @vanatteveldt.com on social media data donation and digital tracking. A very useful chapter on how digital traces can be integrated into social science research, not as a replacement for surveys and other methods, but as a powerful complement.
Taha Yasseri@tahayasseri.bsky.social · 6mo agoChapter 9: Kiran Garimella on using WhatsApp data for computational social science. An important chapter on why WhatsApp deserves far more attention in CSS, not only because of its scale, but because it opens a window onto digital life beyond the usual Western, open-platform focus.
Join us on Thursday, 30 April, at 14:30 CET for the International Roundtable on Computational Social Science with @tahayasseri.bsky.social 🔹 A New Sociology of Humans and Machines 🔹 More info: liu.se/en/event/int...
International Roundtable on Computational Social Science: Taha Yasseri
"A new sociology for humans and machines" Welcome to the International Roundtable on Computational Social Science with Taha Yasseri, Trinity College Dublin. The seminar is open to the public. Language...
liu.se
We are delighted to announce an ONLINE Information Session for our MPhil in Race, Ethnicity and Conflict. 💫 📅 Thurs, 23 April ⏰ 3pm 💻 via Zoom Register via: tcd-ie.zoom.us/meeting/regi... @tcdschoolssp.bsky.social
Chapter 9: Kiran Garimella on using WhatsApp data for computational social science. An important chapter on why WhatsApp deserves far more attention in CSS, not only because of its scale, but because it opens a window onto digital life beyond the usual Western, open-platform focus.
Taha Yasseri@tahayasseri.bsky.social · 6mo agoChpater 8: @dirkhovy.bsky.social, M Gerondeau & J Globisz on text data and natural language processing. A very useful chapter on why text is such a rich source for CSS, and how NLP can help with exploration, prediction, and generation; if used thoughtfully and with clear research goals.