My latest article for the Bulletin of the Atomic Scientists examines how social media platforms and algorithmic curation are fundamentally fracturing society.
The online information landscape, driven in large part by social media, rewards engagement and is curated by classification algorithms. This simple combination is the problem at the heart of society’s fracturing and discord. Societal fragmentation isn’t accidental – it’s structural. Algorithms prioritizing engagement naturally amplify divisive content because controversy drives clicks and interaction. Understanding the fundamental nature of this problem is prerequisite to developing meaningful solutions.
I was a featured expert at NATO’s academic conference on societal resilience in the digital age, held in London. The conference gathered leading academics, defense strategists, and technologists to address how democratic societies can confront the rising threats of disinformation, cognitive warfare, and the strategic use of emerging technologies.
I presented in Session II: The Impact of Social Media on the Resilience of Society, which focused on the dual role of digital platforms as tools for civic engagement and vectors for strategic manipulation. Drawing on our lab’s research into human digital twins and AI-powered misinformation detection, I outlined how generative models – such as large language models – can be weaponized in coordinated influence operations, and emphasized the urgency of building resilient information ecosystems and promoting ethical, interdisciplinary strategies to defend democratic discourse.
We published VirTLab: Augmented Intelligence for Modeling and Evaluating Human-AI Teaming Through Agent Interactions at HCI International 2025 and presented VIRT-LAB: An AI-Powered System for Flexible, Customizable, and Large-scale Team Simulations at UIST 2025.
VirTLab is a system that enables flexible, customizable, and large-scale team simulations using LLM-powered agents. It provides interactive 2D environments for studying human-AI dynamics, allowing researchers to model and evaluate how humans and AI agents can work together in team settings.
This work is part of our broader research into Human-AI Teaming (HAT) funded by DARPA, exploring how we can better understand and improve collaboration between humans and AI systems.
Co-authors: Mohammed Almutairi, Charles Chiang, Haoze Guo, Matthew Belcher, Nandini Banerjee, Maria Milkowski, Svitlana Volkova, Daniel Nguyen, Tim Weninger, Michael G. Yankoski, Diego Gomez-Zara