Why ESL Researchers Should Think Twice Before Using GPT to Edit a Manuscript?
Working with GPT for the polishing of a manuscript has associated risks. It can create problems at the stage of journal submission even if one uses it exclusively to improve language quality.
Many journals and publishers now have policies related to generative AI, and some use tools to screen manuscripts in an attempt to detect AI-generated content. The results of such screening can lead to different consequences, including requests for clarification, disclosure of AI use, closer scrutiny, rejection, or other actions, depending on the policy of the specific journal.
It creates an interesting dilemma for the ESL researcher.
On the one hand, you can resort to GPT to improve your English, but at the same time, the final version of the manuscript will not necessarily reflect your own writing process anymore. In cases where the research itself is completely done by you, extensive rewriting by AI can raise questions about the nature and extent of the assistance used.
It is critical to differentiate between applying a tool to perform only minimal language correction and using the same tool to generate large sections of a manuscript.
Therefore, researchers cannot assume that their use of GPT to edit their paper is harmless simply because they conducted the original research themselves.
Questions to Ask Before Using GPT to Edit a Manuscript
When a researcher considers using GPT to edit a manuscript for publication, they need to ask:
- Is generative AI-assisted editing allowed by the target journal?
- Does the journal require disclosure of any use of AI?
- Does the journal draw a distinction between proofreading and rephrasing?
- Can extensive rewriting create questions about authorship or originality?
- Can uploading the manuscript create issues related to confidentiality and intellectual property?
- How does the journal handle cases when an editor inquires about the use of AI?
The most important consideration in any case is that one cannot assume that an AI detector will reliably recognize AI assistance. These tools are not perfect and can produce both false negatives and false positives.
However, compliance with the journal's policy remains essential.
AI Detection Is Not a Substitute for Following Journal Policies
Researchers should not make decisions about AI-assisted editing based solely on whether an AI detector is likely to identify the use of an AI system.
AI detection systems can have limitations, and their results should not be treated as definitive evidence on their own. More importantly, a journal's rules about generative AI use apply regardless of whether an automated detection system identifies the assistance.
The relevant question is therefore not simply whether a journal can detect AI assistance. Researchers should first determine what the journal permits, what it requires authors to disclose, and whether their intended use falls within those requirements.
Why ESL Researchers May Want to Avoid Extensive AI Rewriting
Researchers who use GPT to enhance their English proficiency may find it tempting to allow the technology to rewrite large portions of a manuscript.
However, extensive rewriting can create additional questions about who formulated particular parts of the final text and whether the resulting manuscript accurately represents the author's intended meaning.
For ESL researchers, it can therefore be sensible to maintain full control over the manuscript and avoid unnecessary AI rewriting.
If there are problems with the language, a professional human academic editor can provide contextual editing without requiring AI to rewrite the entire manuscript.
Can AI Editing Cause Rejection?
It might, depending on the policies of the journal and the manner in which AI was used.
If a journal determines that a manuscript does not follow its requirements concerning AI use or disclosure, or if the use raises questions under the journal's authorship or originality criteria, the journal may take action under its applicable policies.
Possible consequences depend on the specific circumstances and the journal's rules. They can include requests for clarification, disclosure requirements, additional review, or rejection.
It therefore makes sense to view AI-assisted editing as something that should be considered carefully within the publication process rather than as a risk-free way to improve English.
What About “Hall of Shame”?
When researchers are warned about the possibility of being publicly shamed as an example of improper AI use, this should not be treated as a standard feature of scientific publishing.
Editors and publishers of scientific journals generally have policies for responding to suspected research misconduct or publication concerns.
Depending on the circumstances and the applicable policies, corrective measures can include correction statements, rejection, expressions of concern, or communication with the authors' institutions.
The appropriate response depends on the specific facts of the case, the journal's policies, and the nature of the concern.
Take-Away Message
Do not count on using GPT to edit your paper simply because it seems like an easy route to perfect English. Rather, begin by considering the possible implications for publication.
The research might be novel, the data authentic, and the conclusions genuinely yours. But incorrect or unauthorized use of AI tools could create another ethical or publication-related problem.
An ESL author's priority should be preserving the paper's integrity and ensuring compliance with the requirements of the target journal, rather than simply polishing the language.
If you need help improving the language of your manuscript while retaining control over its scientific meaning, consider a contextual human academic editing review.