NewsBot: AI for News Audiences
News organizations are deciding how to bring large language models into production, distribution and audience engagement. Before building a news chatbot, we asked how journalists and readers want to use one.
Journalists preferred a human in the loop, especially for subjective questions. Readers asked a chatbot more factual questions than they asked authors, and questioned journalistic integrity more often with authors. We also help build multilingual models for news analysis, such as LlamaLens.
ResourcesQuestions asked to authors vs. a chatbot (dataset)
Publications
-
NAACL Findings 2025
LlamaLens: Specialized multilingual llm for analyzing news and social media content
Findings of the Association for Computational Linguistics: NAACL 2025
-
CHI 2024
Towards Designing a Question-Answering Chatbot for Online News: Understanding Questions and Perspectives
Proceedings of the 2024 CHI Conference on Human Factors in Computing Systems