Philine Widmer's research lies at the interface between political economy, data science, and artificial intelligence. She studies traditional news media, social media and their recommender algorithms, and artificial intelligence systems such as large language models, asking how these relate to political and market institutions. She uses field and survey experiments as well as large-scale observational data. One of her most prominent projects shows that the feed algorithm of X shifts users' political opinions in a more conservative direction (published in Nature in 2026).
A second strand of her research develops methods for unstructured data use, especially in the social sciences. Much of what matters politically is recorded as text, images, audio, and video rather than as numbers, and she builds econometric and machine-learning tools that make such material amenable to statistical inference. These methods can recover the narratives running through a corpus of news, or quantify the ideological position taken by a political advertisement.
Research axes
Economics of news media, social media, and recommender algorithms
Political economy of artificial intelligence and large language models
Applied microeconometrics and machine learning for unstructured data (text, images, audio, video)
