My awesome co-author Diliara Valeeva describing our work simulating social media using LLMs which is short-listed for a French innovation award!

Researchers at the University of Amsterdam used hundreds of AI personas to find the best model for social media news feeds in terms of toxicity and bridging the political divide.

Petter Törnberg Justus Uitermark Chris Bail (chris_bail_duke 🧵) #GlobalTrendsForum

⏰📅 Only 3 days left to apply for the Summer Institute in Computational Social Science at Università di Bologna (deadline: April 25, 12 pm, GMT+1). Don't miss out on this free opportunity to join a unique program! Apply now: sicss.io/2024/bologna/ #SICSS #CSS

Ever wish your research *did not* replicate? Our study that suggests Americans become *more* polarized when they are exposed to a common enemies (not less) may generalize to Korea and Israel: shorturl.at/dlnCF (original study here: shorturl.at/almT8)

What a beautiful Essay on 'Re-wilding' the internet, by Maria Ferrell and Robin Berjon | 🦋 robin.berjon.com : noemamag.com/we-need-to-rew…

The future of AI is open source- latest Llama model compares very favorably with two of the latest and greatest proprietary models. Still not completely open, but an exciting step forward! ai.meta.com/blog/meta-llam…

Today my wife Nina Guilbeault, PhD launches her timely and important book THE GOOD EATER w. Bloomsbury U.S.! It’s an incredibly thoughtful and open-minded analysis of our food system and how you, as a consumer and eater, can navigate its many trade-offs while staying aligned with your

'Because algorithms reveal more of their biases, people are also more likely to correct their biases when decisions are attributed to an algorithm than to themselves.' pnas.org/doi/epdf/10.10…

'Humans had no natural predators, and then we created AI... it's like if salmon invented grizzly bears'

Researchers at Stanford pitted AI against Human Experts in a war-game simulating conflict with China: arxiv.org/pdf/2403.03407…

New experiment shows that people who interact with an AI chatbot vs. a person report 'feeling heard' more often--yet the effect disappears when the people learn they are interacting with an AI chatbot: shorturl.at/cnuwx. Surprisingly, no one was asked to leave their wife.

Youtube, Twitter, and LinkedIn usage are all trending downward over the last year according to the latest USC Neely center survey on social media usage. Reddit, on the other hand, appears to be growing.

New post-doc at Data and Society: boards.greenhouse.io/datasocietyres…

'Our results suggest that between 6.5% and 16.9% of text submitted as peer reviews to these conferences could have been substantially modified by LLMs': arxiv.org/pdf/2403.07183…

Sciences Po Paris is hiring an Associate Professor: medialab.sciencespo.fr/static/PostePu…

Cool new paper uses LLMs to correct misinformation by Tim Althoff, Amy Zhang, and others: arxiv.org/pdf/2403.11169…

Still time to apply to SICSS-Rutgers! June 17-28 in New Brunswick, NJ. Free training in computational social science for doctoral students, postdocs, and faculty. sicss.io/2024/rutgers/