Have you ever wondered how the writing styles of AI and human journalists compare? A study from Charles Darwin University (CDU) takes a closer look at this intriguing topic. Published in the journal Computers, the research dives into the unique linguistic traits that set AI-generated articles apart from those crafted by humans.
Researchers at CDU examined 150 articles from well-known publications like the New York Times and the Australian, stacking them up against 150 pieces created by the AI chatbot Gemini. Van Hieu Tran, the lead investigator, points out that although AI content is getting more polished, it still lacks the creative flair and diversity that human journalists bring to the table.
Human writers tend to vary their sentence and paragraph lengths more, and they use a wider array of verbs, which shows their focus on action and engagement. On the flip side, AI-generated content often leans heavily on nouns and can come across as syntactically repetitive.
Tran puts it simply: “Our research shows that both AI and human writers produce content that’s easy to read. However, human writers offer more diverse sentence structures and paragraph styles, which AI hasn’t quite mastered yet. AI content can seem a bit ‘boring’ because it lacks that stylistic diversity and personal touch.”
Dr. Yakub Sebastian, a co-author of the study, emphasizes the value of human creativity in journalism. He notes the importance of distinguishing between AI and human writers, especially considering AI’s potential biases and its ever-evolving capabilities.
“AI models are advancing rapidly, and they’re getting better at mimicking human tasks,” Dr. Sebastian observes. With the launch of entirely AI-generated newspapers in Italy, identifying AI-created content is becoming more challenging.
To tackle this issue, the researchers suggest developing tools similar to plagiarism detectors that can spot AI-generated news. Such innovation would help maintain transparency and uphold the integrity of journalistic work.
