The ability of chatbots, like ChatGPT, Claude, and Gemini, among many others [1] to generate human-like language in response to a wide range of prompts makes them a tempting tool for authors. Online videos describe how to use ChatGPT to craft titles, abstracts, elements of a paper’s introduction and background (e.g., the literature review), and even the results and discussion sections (try searching for “using ChatGPT to write a scientific paper” in YouTube).
Because chatbots are useful, I suspect that this kind of use – the crafting of de novo text – is growing. As explained below, de novo content generation by chatbots can be problematic. But some chatbots can also revise and clarify and journals should promote this kind of use.
Chatbot challenges
Concerns about the “original” chatbot, ChatGPT, quickly arose among publishers of academic journals following the tool’s November 2022 launch. The World Association of Medical Editors explained in May 2023 that “… chatbot output currently carries the risk of including biases, distortions, irrelevancies, misrepresentations, and plagiarism…” In concept, holding authors accountable for chatbot-generated content should address the potential for bias, distortion, irrelevancies, and misrepresentations (in the same way that this accountability addresses these issues in the context of author-written text).
The potential introduction of plagiarism by chatbots poses a distinct challenge, however. For example, because ChatGPT’s output draws on existing content, much of it may be plagiarized.[2] Holden Thorpe, the editor-in-chief for Science, has gone further than prohibiting re-use only material that a chatbot draws from other sources and has instead seemed to rule out all chatbot-generated text. He explained, “For the Science journals, the word ‘original’ is enough to signal that text written by ChatGPT is not acceptable: It is, after all, plagiarized from ChatGPT.”
The impact of journal policy on chatbot use
Academic journals (including Science) have settled on an intermediate position. The International Committee of Medical Journal Editors [3], World Association of Medical Editors [4], and Science Journals [5] do not permit designation of a chatbot as an author (because, for example, a chatbot cannot approve its output for publication or understand a conflict of interest [6]), but do permit its use if disclosed.
Unfortunately, journal policies may nonetheless be discouraging chatbot uptake. I am unaware of data on use of these tools use by academic authors, but my own peer review experience suggests far less than universal use for stylistic editing. Moreover, if authors are using chatbots, they do not seem to be reporting it. For example, none of the 16 articles published in the April 2024 issue of Value in Health acknowledges using a chatbot.
The benefits of promoting chatbot use
While journals have appropriately crafted standards for responsibly using chatbots, they should also recognize the potential benefits and encourage their use to stylistically revise and improve author-composed text. This use of chatbots – in contrast to de novo text generation – should mitigate plagiarism concerns.
Crucially, encouraging authors to use chatbots has the potential to make scholarly publication more accessible. A study [7] of professionals found that ChatGPT improved the quality of their writing, helping authors with the lowest baseline writing skills improve the most. Commenting on these findings, Robert Seamans, an economist on the faculty at New York University expressed the opinion that “[ChatGPT] has almost like a democratizing effect.” Second, as the same study reported, ChatGPT makes writers more productive. Finally, better writing makes scholarly manuscripts easier for peer reviewers to evaluate and for readers to understand.
To be sure, chatbots will not ensure that writing makes sense, that methods are valid, or that conclusions are well-supported. But by saving time, they can reduce the “cost” of publishing scholarly work while improving its clarity and hence its reach, influence, and value. Journals should move away from policies that can seem to grudgingly permit chatbot use and instead enthusiastically promote the appropriate and responsible incorporation of chatbots into the author’s toolbox.
[1] As this article suggests, chatbots have a range of different features, and some may be more appropriate than others for the application I discuss here.
[2] “A new report from plagiarism detector Copyleaks found that 60% of OpenAI's GPT-3.5 outputs contained some form of plagiarism.”
[3] Work may not credit a chatbot as an author: “Chatbots (such as ChatGPT) should not be listed as authors because they cannot be responsible for the accuracy, integrity, and originality of the work, and these responsibilities are required for authorship”. Work must disclose use of AI to assist with writing: “At submission, the journal should require authors to disclose whether they used artificial intelligence (AI)-assisted technologies (such as Large Language Models [LLMs], chatbots, or image creators) in the production of submitted work. Authors who use such technology should describe, in both the cover letter and the submitted work in the appropriate section if applicable, how they used it. For example, if AI was used for writing assistance, describe this in the acknowledgment section…”
[4] Work may not credit a chatbot as an author: “Chatbots cannot be authors… under most jurisdictions, an author must be a legal person. Chatbots do not meet the International Committee of Medical Journal Editors (ICMJE) authorship criteria, particularly that of being able to give “final approval of the version to be published” and “to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.” (10) No AI tool can “understand” a conflict-of-interest statement and does not have the legal standing to sign a statement. Chatbots have no affiliation independent of their developers. Since authors submitting a manuscript must ensure that all those named as authors meet the authorship criteria, chatbots cannot be included as authors.” Work must disclose use of AI to assist with writing: “WAME Recommendations 2.1: Authors submitting a paper in which a chatbot/AI was used to draft new text should note such use in the acknowledgment; all prompts used to generate new text, or to convert text or text prompts into tables or illustrations, should be specified.”
[5] Work may not credit a chatbot as an author: “AI-assisted technologies [such as large language models (LLMs), chatbots, and image creators] do not meet the Science journals’ criteria for authorship and therefore may not be listed as authors”. Work must disclose use of AI to assist with writing: “Authors who use AI-assisted technologies as components of their research study or as aids in the writing or presentation of the manuscript should note this in the cover letter and in the acknowledgments section of the manuscript.”
[6] See preceding three notes.
[7] See Figure 2A. Both groups submitted a writing assignment without any assistance to establish a baseline writing score (Task 1). The treatment group (triangles) received ChatGPT assistance for Task 2, but the control group (squares) did not. Note that for Task 2 (vertical axis score) the treatment group grade (triangles) exceeds the corresponding control group’s Task 2 grade (squares), thus indicating that ChatGPT improved the quality of the writing. Note also that the gap between the treatment group and control group is greater for participants with low Task 1 grades (horizontal axis score).
