AI chatbot accountability in the age of algorithmic gatekeeping: Comparing generative search engine political information retrieval across five languages

This study investigates the performance of search engine chatbots powered by large language models in generative political information retrieval. Applying algorithmic accountability as a central theme, this research (a) assesses the alignment of artificial intelligence (AI) chatbot responses with ti...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:New media & society
Hauptverfasser: Kuai, Joanne, Brantner, Cornelia, Karlsson, Michael, Van Couvering, Elizabeth, Romano, Salvatore
Format: Journal Article
Sprache:Englisch
Veröffentlicht: 28.02.2025
Schlagworte:
AI-mediated communication
algorithm audit
algorithmic accountability
digital infrastructure
election news
gatekeeping
GenAI chatbot
generative AI
Media and Communication Studies
Medie- och kommunikationsvetenskap
Microsoft Copilot
quantitative content analysis
search engine
ISSN:1461-4448, 1461-7315, 1461-7315
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:This study investigates the performance of search engine chatbots powered by large language models in generative political information retrieval. Applying algorithmic accountability as a central theme, this research (a) assesses the alignment of artificial intelligence (AI) chatbot responses with timely political information, (b) investigates the factual correctness and transparency of chatbot-sourced synopses, (c) examines the adherence of chatbots to democratic norms and impartiality ideals, (d) analyzes the sourcing and attribution behaviors of the chatbots, and (e) explores the universality of chatbot gatekeeping across different languages. Using the 2024 Taiwan presidential election as a case study and prompting as a method, the study audits responses from Microsoft Copilot in five languages. The findings reveal significant discrepancies in information readiness, content accuracy, norm adherence, source usage, and attribution behavior across languages. These results underscore the contextual awareness when applying accountability assessment that looks beyond transparency in AI-mediated communication, especially during politically sensitive events.
ISSN:1461-4448
1461-7315
1461-7315
DOI:10.1177/14614448251321162