Human-written articles easier to grasp than AI-written ones
Originally posted on The Horizons Tracker.
News articles crafted by humans are easier to understand than those produced by automation, according to a study1 by Ludwig Maximilian University of Munich (LMU). Researchers in the Department of Media and Communication surveyed over 3,000 online news readers in the UK, each rating one of 24 articles—half written by journalists, half created with AI assistance.
“Readers found the 12 AI-produced articles significantly less comprehensible,” the study notes, despite these pieces having been edited by journalists before publication.
Struggling to understand
The findings suggest that readers struggled with the language in AI-generated texts, often encountering inappropriate, complex, or uncommon words. Additionally, readers found the AI articles less effective in presenting numbers and data, which made them harder to follow. However, both AI and human-written articles received similar ratings for narrative structure and flow.
The researchers recommend that journalists and technologists focus on simplifying numerical data, explaining unfamiliar terms, and incorporating language that helps readers visualize the story when preparing automated content.
This study is the first to examine the readability of human versus AI-generated news and explore the reasons behind the difference. The findings highlight the need not only for human oversight in automated journalism but also for refining how data-driven stories are produced.
Article source: Human-Written Articles Easier To Grasp Than AI-Written Ones.
Header image source: Kaboompics.com on Pexels.
Reference:
- Thäsler-Kordonouri, S., Thurman, N., Schwertberger, U., & Stalph, F. (2024). Too many numbers and worse word choice: Why readers find data-driven news articles produced with automation harder to understand. Journalism, 14648849241262204. ↩
