E. M. Wolkovich was accused of producing ChatGPT text.
I have been accused of committing scientific fraud, but not data fraud. No one has accused me of fabricating or manipulating data or results. This comes as a relief because my laboratory focuses on understanding how global change affects ecological communities, and we take great care to ensure that our data is transparent, shareable and reproducible. Instead, I have been accused of committing writing fraud by passing off content produced by AI as my own. This accusation is difficult for me to accept because, like many people, I find writing to be a challenging process. I often read books on writing to find ways to improve and comfort myself with the knowledge that writing is generally slow and difficult. As a scientist, I believe it is important to communicate effectively and always advise my students to create outlines before drafting and revising their work.
I was taken aback when I received feedback from reviewers on a submitted paper who claimed that it was written by ChatGPT. One reviewer even stated that it was “obviously ChatGPT,” and the handling editor found the writing style to be unusual. I felt a range of emotions, including surprise, shock, dismay, and confusion. It was a blow to be accused of being a chatbot without any evidence, especially after putting so much work into the writing process.
In reality, I did not use ChatGPT to write a single word of the manuscript. I quickly tried to think of ways to prove my case. Since I use plain-text files and track my progress using the version-control system Git, I was able to show the text change history on GitHub, including commit messages like “finally writing!” and “Another 25 minutes of writing progress!” that I never thought I’d have to share. I also compared the writing style of my previous papers to the current submission.
I considered asking ChatGPT itself if it had written my paper, but ultimately decided against it. I didn’t want to spend my time proving that I’m not a chatbot; I wanted to find perspectives from other researchers who work on data fraud, co-authors on the paper, and colleagues. Most of the people I spoke with agreed with my alarm and one person summed it up by saying “All scientific criticism is admissible, but this is a different matter.”