The availability of artificial intelligence-powered research paper mills may threaten research integrity in the modern age.
The latest version of OpenAI’s artificial intelligence (AI) chatbot, ChatGPT, has impressed many with its ability to produce poetry and academic essays that are hard to distinguish from those written by humans. However, some are concerned that this could lead to the use of AI for research fraud and a decrease in the quality of research output and publication. There are worries that easily accessible and unreliable linguistic content could overshadow original research that requires significant effort and resources.
“If an unreliable linguistic mash-up is freely accessible, while original research is costly and laborious, the former will thrive,” warned The Financial Times associate editor and chief business commentator John Gapper.
The availability of AI-powered research paper mills posing a threat to research integrity is not a theoretical concept. It is already happening.
Ethical and integrity concerns are becoming increasingly prevalent in academic publishing. F1000 found 34% of their articles and research outputs were subject to ethics cases, while 50% of Taylor & Francis’ research articles and scholarly papers had ethics issues. Both these publishing platforms had to deal with ethics issues such as duplicate submissions, data integrity, citation manipulation, and authorship integrity.
In the biomedical literature field, Dutch microbiologist and scientific integrity consultant Elisabeth Bik drew attention to the fact that AI was contributing to image fraud using Generative Adversarial Network (GAN) technology, which is capable of producing deep fakes. These examples prove that the magnitude of this issue is significant due to the various types of paper mills that exist and their high adaptability.
To help combat this predicament, publishers have to play a crucial role in ensuring that the scholarly content published and disseminated is legitimate and upholds integrity. They can invest in various systems, safeguards, and expertise to ensure proper procedures are followed. It is also necessary to have scalable tools to identify research integrity and publishing ethics issues.
Open research is another essential aspect to help reduce research waste and enable the scrutiny of data, which was widely agreed upon during a recent Westminster Higher Education Forum. As AI and automation become increasingly integral to research – particularly in big-data analysis – making data open will significantly benefit the use of AI in identifying fraudulent data. In addition, promoting and enabling the publication of a wide range of outputs, including negative and null findings, protocols, and incremental studies – a vital element of the open research model – helps reduce publication and editorial bias. It also provides additional accessible data for AI tools to detect research fraud.
Publishers must also be willing to collaborate with stakeholders, such as developers, other publishers, and scholarly organizations, to address the root causes of research fraud. A reward and incentive system that may encourage the use of paper mills could be an option.
The STM Integrity Hub and its prototype paper mill detector demonstrate the benefits of cross-publisher collaboration. Automated AI processes that detect duplicate publications and other issues between publishers are crucial, given that publishers often have distinct submission and publication systems.
Although AI tools are essential for detecting research integrity and publishing ethics issues, human judgment also plays a crucial role in ensuring research integrity. Rigorous checks conducted by AI and experienced experts before publication are necessary to maintain research integrity in publications.
To safeguard research integrity and prevent misconduct in scholarly publishing, it is also important to provide researchers with high-quality education and training in publishing ethics. Good research practices, such as highlighting authors’ responsibilities, constitute a good peer review.
While the future of research integrity may seem bleak in the face of AI technology, it is not predetermined. Research integrity is shaped by human actions and how we respond to situations.
Do you think AI will be a threat to academic publishers in terms of research and publishing? Share with us your thoughts in the comments below!
This article is published in collaboration with Social Science Space, an online social network that brings social scientists together to explore, share, and shape the significant issues in social science – from funding to impact. To read the full original text, click here.