Research · Updated 26 Oct, 11:20 pm IST

Researchers identify six linguistic patterns that make AI text tiring to read

Why it matters for readers: It shows why AI writing sometimes feels awkward and what researchers are doing to measure those problems.

The study analyzed 120 conversations, 45 coding sessions and 980 essays to compare model text with human writing.1
Researchers found six overused linguistic constructions in AI text, such as unusual word combinations and long-distance sentence dependencies that tax readers' working memory.1
Common problems included dropped connectors (e.g., 'the book the professor wrote'), stacked nouns, turning nouns into verbs, and placing related sentence parts far apart.1
The dataset included outputs from Claude, GPT-5.6, GLM-5, Kimi K2.5 and smaller open-source models like Llama and Olmo.1
Examples from the study show models producing awkward phrasings such as 'A cat who can always get up and away is a much calmer cat,' illustrating long-distance separation of subject and predicate.1

Get a brief like this every morning

Uzha reads hundreds of sources and gives you the stories that matter for your work, with every source linked. Free.