Our study shows that coordination naturally emerges in social simulations with LLM-based agents. When a small group shares a goal and awareness of teammates, they develop human-like coordination strategies without explicit human guidance.

Luca Luceri@luceriluc.bsky.social · 9mo ago

🚨 New preprint 🚨 Can AI agents coordinate influence campaigns without human guidance? And how does coordination arise among AI agents? In our latest research, we simulate LLM-powered AI agents acting like users on an online platform, some benign, some running an influence operation

The first large-scale investigation of multimodal AIGC on social media! 📄Pre-print: arxiv.org/abs/2502.11248 📷We made our resources available, including 2M+ labeled images - visit: github.com/angelayejinyi/AIGC-Election-2024

Zhiyi Chen@zhiyi-chen.bsky.social · last yr.

🚀 First study on multimodal AI-generated content (AIGC) on social media! TLDR: AI-generated images are 10× more prevalent than AI-generated text! Just 3% of text spreaders and 10% of image spreaders drive 80% of AIGC diffusion, with premium & bot accounts playing a key role🤖📢

Are your friends more susceptible to influence than you? Our new work in #ICWSM2025 uncovers The Susceptibility Paradox in online social influence. w/ @luceriluc.bsky.social , Julie Jiang, @emilioferrara.bsky.social Explore our findings: arxiv.org/abs/2406.11553

A rent paper has been portrayed in the news as proof that Twitter/X new #recsys favors republicans. In our latest work, collecting millions of datapoints via 120 sock puppets, we should the reality is much more complex: you would be surprised by our findings: arxiv.org/abs/2411.01852 thread 1/n

Auditing Political Exposure Bias: Algorithmic Amplification on Twitter/X Approaching the 2024 U.S. Presidential Election

Approximately 50% of tweets in X's user timelines are personalized recommendations from accounts they do not follow. This raises a critical question: what political content are users exposed to beyond...

arxiv.org