DATATALES: Investigating the use of Large Language Models for Authoring Data-Driven Articles

“The emergence of contemporary large language models (LLMs) and their remarkable text generation capabilities led to increased interest in assessing their value for a range of creative writing tasks [7], including data storytelling [18]. While this technology has the potential to fundamentally reshape the way people use writing tools [31], it also introduces news challenges such as unreliable outcomes, lack of domain understanding, prompt complexity, ethical concerns, among others [18]. We believe that these issues require thoughtful design solutions to circumvent them, and that different writing genres may benefit from purpose-specific features built around these models.

In this work, we investigate the potential of LLMs to support the authoring of data-driven articles. Based on the deep intertwining of charts and text in these articles, and targeting the intermediate stages of the visual storytelling process where authors are actively building a story based on exploratory findings [16], we propose chart interaction as a more intuitive alternative to direct prompting for conveying narrative intent to the LLM. We developed an early proof-of-concept, DATATALES, that generates textual content for an accompanying chart, and additionally allows authors add chart annotations to guide focus of the story. Authors can use the generated text as-is, edit portions of the text, or generate multiple instances to pick-and-choose what they like.”