In recent months, the scientific community has been increasingly discussing the role of artificial intelligence tools, such as ChatGPT, in the process of producing scientific publications. Numerous cases of undisclosed use of AI are emerging, raising concerns about the integrity and transparency of research. 

Scale of the problem 

Analyses conducted by Prof. Artur Strzelecki of the University of Economics in Katowice showed the presence of characteristic phrases generated by ChatGPT in many peer-reviewed scientific articles, including prestigious journals indexed in the Scopus database. Phrases such as “As an AI language model” or “As of my last knowledge update” appeared in texts without proper labeling, suggesting unconscious or intentional copying of AI-generated content. In some cases, authors did not even remove the “regenerate response” button, indicating a lack of thorough verification in the editorial process. 

Reactions from publishers and the scientific community 

In response to growing concerns, scientific publishers such as Springer Nature and Taylor & Francis have introduced guidelines for the use of AI in publications. They emphasize the need to disclose any use of AI tools in the article writing process, especially when they affect the substantive content. At the same time, they point out that AI cannot be considered a co-author of a publication, as it does not meet the criteria of responsibility for scientific content. 

Challenges of AI detection 

Detecting undisclosed use of AI in publications is a major challenge. Studies show that even experienced reviewers have difficulty distinguishing texts written by humans from those generated by AI. Additionally, AI can create false or non-existent citations, leading to the dissemination of unverified information in scientific literature. 

Need for education and transparency 

Many experts, including Prof. Strzelecki, emphasize the need to educate scientists on the responsible use of AI tools. They recommend holding training courses and workshops to help understand the capabilities and limitations of these technologies. Transparency in disclosing the use of AI is key to maintaining trust in scientific publications and ensuring their integrity. 

Sources: naukawpolsce.pl