Faculty Fellow

Biography

Kiran Garimella is a computational social science researcher whose work uses large-scale digital data to tackle societal challenges such as misinformation, political polarization, hate speech, and human migration. He combines theory-driven models from computer science with interdisciplinary collaborations in political science, demography, and journalism to understand how information spreads and how digital platforms can be redesigned for social good. Garimella develops end-to-end research pipelines for collecting, analyzing, and sharing digital trace data.

His projects include measuring and mitigating polarization on social media, studying misinformation and extremism on encrypted messaging platforms such as WhatsApp and Telegram, and using social media and platform data to study migration and other demographic processes. His work appears in venues such as the International AAAI Conference on Web and Social Media, the ACM Web Conference, Proceedings of the National Academy of Sciences, Nature Human Behaviour, Nature Communications, World Development, and Journal of Politics, among others.

Garimella’s research has been supported by the National Science Foundation, the Knight Foundation, the Omidyar Network, Google, and other funders. He received the Adamic-Glance Distinguished Young Researcher Award in 2024, as well as best paper or best student paper awards at ICWSM, WSDM, and WebSci. Before joining Rutgers, he was the Michael Hammer Postdoctoral Fellow at MIT’s Institute for Data, Systems, and Society and a postdoctoral researcher at École Polytechnique Fédérale de Lausanne, following research positions at Yahoo Research Barcelona and the Qatar Computing Research Institute.

Fellowship Project

Young Americans are starting to ask chatbots the questions they used to type into Google. We propose to measure that shift directly, not through self-reports. We will recruit 500-1,000 US adults, oversampling ages 18-29, on Prolific or CloudResearch, and ask them to donate their ChatGPT, Gemini, Google Search, and YouTube histories using data donation tools we have built and deployed at scale (Garimella & Chauchard, 2025; Garimella, 2025). Because these exports reach back months or years, we can observe within each person when, and for which questions, information seeking moved to generative AI.  Our focus is political and civic information: what share of politics-related queries now goes to chatbots, who the early movers are, and what they are told. Search engines and YouTube serve content from a shared inventory that can be audited; a chatbot composes a new answer for every user. We will compare responses different participants received to similar political questions to quantify this fragmentation, building on our analysis of ChatGPT logs donated by 1,300 users in four countries (Roy Chowdhuri & Garimella, 2026) and our audit of citation bias in LLM-based search (Zhang et al., 2027).

The project is built so students can work on every stage of it. Graduate and undergraduate students at SC&I will help design the survey instrument, run the recruitment and consent pipeline, and build the classifiers that identify political queries in donated histories, through research assistantships in my lab, for-credit independent studies, and honors theses. Working with donated behavioral data gives students unusually direct training in research ethics: consent, de-identification, and secure handling of sensitive personal data are part of the daily work, not an abstraction. I will also bring aggregated, de-identified versions of the data into my SC&I courses as class projects, and I would welcome Democracy Lab student fellows into the analysis, for example to co-author the public report or build the public dashboard described below.