NewsChat: A Generative AI Conversational Agent to Enhance the Reading Experience of Data-Driven Articles

Data-driven articles have become an effective way to turn complex datasets into clear, engaging stories. However, this approach can produce dense articles filled with complex details and statistics, which may overwhelm readers. Interactive techniques like visualizations and scrollytelling help simplify content, but often lack personalization and adaptability. This study explores the integration of a Generative AI (GenAI) conversational agent in a data-driven article to simplify content and offer personalized explanations. Through an online experiment,
we assessed how the GenAI agent affects the reading experience in this context. Our results reveal that the agent enhances enjoyment, particularly for individuals with limited interest in the topic. However, the agent may negatively impact perceived credibility, especially among those with skepticism towards chatbots. Thematic analysis of user comments revealed both positive perceptions of utility, alongside concerns about risks such as over-reliance, inaccuracies, and distrust in GenAI.