Research
My research sits at the intersection of communication, computational methods, and artificial intelligence. I study how digital platforms shape public discourse and how emerging AI technologies are transforming both the information environment and the practice of social science research itself. Below are the four areas that anchor my current work.
Information Manipulation in Digital Environments
How does propaganda and strategic communication circulate across social media platforms, and how do platform architectures amplify or constrain these dynamics? While much of the existing literature focuses on authoritarian regimes, I am equally interested in how information manipulation operates within democratic contexts, where partisan actors, interest groups, and platform incentives interact to distort public discourse. My work traces how visual and textual narratives reinforce partisan cues, how echo chambers emerge and sustain themselves, and how platform design choices shape what users see, believe, and share.
Computational Methods for Social Science
Computational methods have opened powerful new avenues for studying human behavior at scale, yet a persistent gap remains between the state of the art in computer science and the methodological practices of social science research. I am interested in how we can bridge this gap not simply by applying new tools to old questions, but by developing approaches that offer deeper, more interpretable understandings of data. This means going beyond prediction to explanation, building methods that are transparent and reproducible, and asking how computational tools can help us see patterns and structures in social life that would otherwise remain invisible.
Generative AI for Social Science
Large language models are increasingly used to simulate human behavior, generate synthetic data, and augment research workflows. But how reliable are these simulations? What are the boundaries of validity, and what procedures should researchers follow to ensure rigor? I am interested in the methodological foundations of LLM-based simulation in social science: the problems of calibration, the challenges of validation against real human data, and the open question of where this line of research should go next. The goal is not to replace human subjects but to understand when and how generative models can responsibly complement traditional methods.
Agentic AI for Research
AI-powered agents are beginning to reshape how research is conducted, from literature review and data collection to analysis and writing. These tools promise dramatic gains in efficiency, but they also raise a harder question: if routine intellectual labor can be automated, what becomes more valuable? My perspective is that thinking and insight, the capacity to ask the right questions, to exercise judgment, to know what matters, are now more important than ever. As researchers, we have long cared about many things; perhaps it is time to also ask what we should stop caring about. If a task can be fully automated, was it ever the kind of work that advanced understanding? I am interested in how we design human-in-the-loop research workflows that use agentic AI to handle the incremental while preserving space for the genuinely creative.