The National Institutes of Health (NIH) recently announced a new policy that prohibits peer reviewers for grant applications from using ChatGPT, to assist them in writing their reviews. The policy states that “the use of ChatGPT or any other NLG [Natural Language Generation] system to generate text for peer review reports is considered a form of plagiarism and a violation of the NIH Peer Review Integrity Policy”. The policy also warns that reviewers who are found to have used ChatGPT or similar systems will face sanctions, such as being barred from future review panels or having their grants revoked.
The decision to ban ChatGPT from peer review sparked a heated debate among researchers and reviewers, who have different opinions on the benefits and risks of using NLG systems to aid their work. Some reviewers argue that ChatGPT is a useful tool that can help them save time and effort, as well as improve the quality and consistency of their reviews. They claim that ChatGPT can generate coherent and relevant text based on the keywords and topics they provide, and that they can edit and refine the output to ensure its accuracy and originality. They also point out that ChatGPT is not intended to replace human judgment or expertise, but rather to augment and enhance it.
However, other reviewers believe ChatGPT has no role in peer review and support the NIH’s ban. They contend that ChatGPT poses serious ethical and scientific issues, such as plagiarism, bias, and lack of accountability. They assert that ChatGPT can produce text that is not based on factual evidence or rigorous analysis, but rather on superficial or misleading associations. They also warn that ChatGPT can introduce errors or inaccuracies into the reviews, which may not be detected or corrected by the reviewers. Moreover, they question the credibility and integrity of the reviewers who rely on ChatGPT to generate their reviews, and suggest that they are not fulfilling their responsibilities as peer reviewers.
The debate over ChatGPT and peer review raises important questions about the role and impact of NLG systems in scientific communication and evaluation. While NLG systems have shown remarkable capabilities in generating natural and fluent text for various purposes and domains, they also pose significant challenges and risks for ensuring the quality and reliability of the information they produce. Therefore, it is essential to establish clear and consistent guidelines and standards for the appropriate and responsible use of NLG systems in peer review and other scientific activities.
Whether this stance is permanent, or reviewed with subsequent iterations of various NLGs remains to be seen. However, one of the objections, based on the concept of uploading private and sensitive information to the platform of a third party may go away in the near future when LLMs may be hosted locally.
A nefarious reviewer can always abuse the review system, and ChatGPT is another way that this could happen. However, peer review has always been based on trust and expertise of third parties. If ChatGPT were used in a way that enhanced the reviewer’s trust and expertise, then it’s hard to imagine a future where it won’t be used. For now, however, its use is, shall we say, ‘discouraged’.
For expert assistance with your grant applications, visit GrantDesk.
Researcher.Life is a subscription-based platform that unifies top AI tools and services designed to speed up, simplify, and streamline a researcher’s journey, from reading to writing, submission, promotion and more. Based on over 20 years of experience in academia, Researcher.Life empowers researchers to put their best research forward and move closer to success.
Try for free or sign up for the Researcher.Life All Access Pack, a one-of-a-kind subscription that unlocks full access to an AI academic writing assistant, literature reading app, journal finder, scientific illustration tool, and exclusive discounts on professional services from Editage. Find the best AI tools a researcher needs, all in one place – Get All Access now at just $25 a month or $199 for a year!
