From Browsing Histor y to ChatGPT Conversations: Data Donation as a Method for Studying AI 's Influence on Human Behaviour
Artificial intelligenc e increasingly shapes what news people read, what content algorithms recommend, and what information people seek out through tools like Ch atGPT. Yet studying these influences on human behaviour requires solv ing two interlinked challenges: getting the right data and being able to analyse it reliably. This talk presents data donation - where par ticipants actively share their own digital trace data with researcher s - as a key method for accessing otherwise opaque information enviro nments. Drawing on a series of projects ranging from browser history and TikTok donations to a recent study collecting ChatGPT conversatio n histories, I discuss what we can learn about AI-mediated informatio n exposure and seeking behaviour, and what remains stubbornly hard. A particular focus is the analytical challenge: when the data is messy , personal, and multi-layered/modal, how do we use computational tool s - including AI itself - to make sense of it, and how much can we trust what we measure?
Dr. Felicia Loecherbach< /strong>
Felicia Loecherbach is an Assistant Professor in Polit ical Communication and Journalism at the Amsterdam School of Communic ation Research. Her research focuses on online news consumption, news diversity, and the use of computational methods in the social scienc es. She is particularly interested in how digital information environ ments shape the ways people encounter, understand, and engage with ne ws, using approaches such as digital trace data, data donations, and computational content analysis. She obtained a PhD in Computational P olitical Communication Science at Vrije Universiteit Amsterdam and wa s previously a postdoctoral fellow at the Center for Social Media and Politics at New York University.
Date: 27 May 2026
Time: 15:00
Location: NU 4B43
DESCRIPTION: From Browsing History to ChatGPT Conversations: Data Dona tion as a Method for Studying AI's Influence on Human Behaviour A rtificial intelligence increasingly shapes what news people read, wha t content algorithms recommend, and what information people seek out through tools like ChatGPT. Yet studying these influences on human be haviour requires solving two interlinked challenges: getting the righ t data and being able to analyse it reliably. This talk presents data donation - where participants actively share their own digital trace data with researchers - as a key method for accessing otherwise opaq ue information environments. Drawing on a series of projects ranging from browser history and TikTok donations to a recent study collectin g ChatGPT conversation histories, I discuss what we can learn about A I-mediated information exposure and seeking behaviour, and what remai ns stubbornly hard. A particular focus is the analytical challenge: w hen the data is messy, personal, and multi-layered/modal, how do we u se computational tools - including AI itself - to make sense of it, and how much can we trust what we measure?Dr. Felici a LoecherbachFelicia Loecherbach is an Assistant Professor i n Political Communication and Journalism at the Amsterdam School of C ommunication Research. Her research focuses on online news consumptio n, news diversity, and the use of computational methods in the social sciences. She is particularly interested in how digital information environments shape the ways people encounter, understand, and engage with news, using approaches such as digital trace data, data donation s, and computational content analysis. She obtained a PhD in Computat ional Political Communication Science at Vrije Universiteit Amsterdam and was previously a postdoctoral fellow at the Center for Social Me dia and Politics at New York University.Date: 27 May 2026Time: 15:00Location: NU 4B43