Political Expression in Private Settings: Evidence from a Survey Experiment on AI Chatbots

Research talk on how AI chatbot design and perceived country of origin may influence political expression, self-censorship, and user behavior.

The Digital Governance Research Colloquium at the Hertie School welcomes Prof. Genia Kostka, Professor of Chinese Politics at Freie Universität Berlin, for a presentation on “Political Expression in Private Settings: Evidence from a Survey Experiment on AI Chatbots.”

The talk examines how people express political opinions when interacting with AI-powered chatbots and whether information about a chatbot's country of origin influences their behavior. Drawing on a survey experiment involving approximately 3,000 respondents from four countries, the research compares interactions with three chatbots presented with different origin cues: no country information, China, or the United States.

The study investigates three forms of constrained political expression: topic avoidance, where users refrain from asking political questions; topic denial, where they subsequently deny having asked such questions; and topic dilution, where less sensitive questions are used alongside sensitive ones. The research explores how chatbot design features may shape users' perceptions and willingness to express political views.

Date: November 25, 2026
Time: 12:00–1:00 PM (CET)
Location: Online via Microsoft Teams
Organizer: Centre for Digital Governance, Hertie School
Series: Digital Governance Research Colloquium
Speaker: Prof. Genia Kostka, Professor of Chinese Politics at Freie Universität Berlin
Admission: Registration required
Audience: Researchers, AI governance professionals, social scientists, and people interested in the relationship between AI and political expression
Focus: AI chatbots, political expression, self-censorship, AI governance, human-AI interaction, country-of-origin cues
Research approach: Cross-sectional survey experiment with approximately 3,000 respondents from four countries
Program: Research presentation on how chatbot characteristics may influence political expression