In a democratic society, people need to be normatively informed about current events to take civic action like voting, protesting, and organizing. In today’s AI-mediated information ecosystem, civic information is highly available, yet can be hard to parse and is often unreliable.
In this talk, Aubin Le Quéré outlines a research agenda to tackle the challenge of balancing societally important epistemic truth with individual wellbeing in the age of generative AI. First, the speaker introduces an example of measuring epistemic truth with an automated infrastructure to assess news knowledge longitudinally. Second, through a large-scale audit of Google’s AI Overviews across all U.S. counties, the speaker shows that LLMs generate credible‑sounding summaries even when underlying sources are unreliable, thus obscuring information quality and requiring users to engage in taxing validation work. Finally, Aubin Le Quéré introduces the concept of “Information Wellbeing,” which holds that people should be aware of current events while limiting adverse impacts on mental health. The talk explores avenues for incorporating these insights into technical tools but argues that regulatory approaches are necessary to solve this fundamentally collective problem.
Bio:
Marianne Aubin Le Quéré’s work traces how AI and other emerging technologies impact online news and civic information ecosystems. Aubin Le Quéré’s active areas of research include understanding the impact of increasingly generative AI information ecosystems, researching models for human-centered informedness, and developing tools, frameworks, and guidelines for LLMs as a research method. She leverages mixed-method techniques to drive insights in the fields of social computing, computational social science, and communication. Her research has been published in general science venues and in the proceedings of top computer science conferences, including CHI, CSCW, and ICWSM. Her honors include receiving a Top Paper Award from the International Communication Association in 2024 and being named a 2021 Facebook Fellowship Finalist. Aubin Le Quéré completed internships at Mozilla, Bell Labs, and Reddit during her Ph.D. at Cornell University. Before beginning her doctoral studies, Aubin Le Quéré earned a B.S. in computer science and nonfiction from Brown University and worked as a product manager at Microsoft.
In-person attendance is open to Princeton University faculty, staff and students.
This talk will be livestreamed and recorded. The recording will be posted to the CITP website, the Princeton University Media Central channel and the CITP YouTube channel.
If you need an accommodation for a disability please contact Jean Butcher at butcher@princeton.edu.
Sponsorship of an event does not constitute institutional endorsement of external speakers or views presented.