Modern AI technologies are becoming more and more common in the educational process and in the field of journalism. But to what extent can they be trusted? Are they infallible in determining the authorship of texts? Academic integrity and control over the use of AI are the challenges faced by teachers and researchers.
Researchers from the Department of Social Communications at Mariupol State University, Professors Tetiana Ivanova and Oksana Yefremova, conducted an interesting experiment, the results of which raise serious questions about the reliability of modern AI content detectors. The research consisted of checking the authenticity of texts using popular online resources:
Three versions of the text were used for the analysis:
The results of the test were unexpected: ❌ None of the services could correctly identify the texts' authorship. 🤯 The original text from 2019 was mistakenly identified as 100% AI-generated. 🤷♀️ On the other hand, a fully AI-generated text was recognised as either authorial or only partially generated (37%).
These discoveries highlight the importance of a critical approach to the use of automated tools in teaching and proofreading. "Teachers and journalists should not only control AI, but also understand its capabilities and limitations”, said Tetiana Ivanova, media expert at the Academy of Ukrainian Press, Doctor of Pedagogy, Professor, Head of the Department of Social Communications at Mariupol State University.
Join the training and discussions!
The Academy of Ukrainian Press invites everyone interested in artificial intelligence to a series of webinars to be held in March with the support of the Friedrich-Naumann-Stiftung für die Freiheit. Together with leading experts, we will discuss the challenges and opportunities of AI in journalism and education.
🚀 Don't miss the registration and join the professional discussion!
What resources do you use to check the authenticity of texts? Share your experience in the comments!
