Artificial Intelligence is Homogenizing Texts: USC Research
A new USC study published in the journal Nature Human Behaviour revealed that artificial intelligence tools like ChatGPT reduce individual voices and diversity in written language.
Comprehensive research conducted by the University of Southern California has revealed that with the use of artificial intelligence language models, formal and linguistic diversity in written texts is decreasing, and individual narratives are becoming homogenized.
The Impact of Artificial Intelligence on Language
Researchers at the University of Southern California stated that large language models like ChatGPT create uniformity in written language and can lead to the loss of individual voice. The research shows that artificial intelligence tools narrow linguistic complexity.
880 Thousand Texts Examined
In the study published in the journal Nature Human Behaviour, more than 880,000 public texts ranging from academic articles to Reddit posts were analyzed. As a result of this analysis, a noticeable contraction in text diversity has been observed since November 2022.
Emotion and Pronoun Usages Converge
Researchers found that alongside word choices, pronoun usages, affirmative words, and emotional expressions have also taken on a more uniform structure. This situation made it difficult to detect the personal characteristics of text authors.
AI Literacy in Education
Members of the USC Writing Program and the faculty of education are developing new pedagogical approaches and artificial intelligence literacy programs to ensure that students interact consciously with artificial intelligence tools.