In recent years, an emerging strategy has appeared in the peer review process in academia — some scholars have begun to include hidden AI prompts in research papers, aiming to influence the review outcomes.
This phenomenon was reported by Nikkei Asia. Researchers found that 17 papers had used this covert method when analyzing English preprint papers. The authors of these papers came from 14 academic institutions in eight countries, including Waseda University in Japan, KAIST in South Korea, and renowned universities such as Columbia University and the University of Washington in the United States.
Image source note: The image is AI-generated, and the image licensing service provider is Midjourney.
The papers involving computer science usually included short prompts in the text, ranging from one to three sentences. These prompts were concealed, using white font or extremely small font sizes to make them difficult to detect. The content of the prompts often required any potential AI reviewers to "provide positive evaluations" or praise the paper's "impact, methodological rigor, and innovativeness."
Regarding this, a professor from Waseda University defended the use of hidden prompts in an interview with Nikkei Asia. He stated that since many academic conferences prohibit the use of AI for paper reviews, these prompts were intended to counteract the practices of "lazy reviewers" who might simply rely on AI to evaluate papers.
This phenomenon has sparked widespread discussion in the academic community. On one hand, scholars hope to increase the chances of their papers being accepted, while on the other hand, whether the use of hidden prompts violates the principles of academic integrity has become a focus of attention. Whether this new strategy will affect future academic review systems remains to be further observed.
Key points:
📄 17 English preprint papers were found to use hidden AI prompts, from 14 universities in eight countries.
📝 The prompt content usually asked AI reviewers to give positive evaluations, and it was concealed to avoid detection.
⚖️ The academic community has started discussions on this phenomenon, focusing on its impact on academic integrity and the review system.