Generative AI must not be used to directly generate application materials, generative AI must not be listed as a co-completer of the results, and researchers must not disseminate research results to the public without scientific verification or peer review......
Recently, the Ministry of Science and Technology issued the Guidelines on the Code of Conduct for Responsible Research (2023) (hereinafter referred to as the "Guidelines"), which clearly draws red lines for hot issues such as artificial intelligence and the release of major achievements that are of concern to the society.
"Technological progress has led to changes in the research environment. The international community is also strengthening the integrity of scientific research and the construction of science and technology ethics, and it is imperative to revise and improve the academic norms of the mainland. As an expert involved in the drafting and revision of the document, Yuan Junpeng, a researcher at the Library and Information Center of the Chinese Academy of Sciences, said in an interview that under the new situation, it is of great significance to clarify what is right and what cannot be done in scientific research, and to put forward an implementable and operable specification for scientific research institutions, scientific research personnel and other subjects.
Compared with the previous academic style documents issued by the mainland, what are the innovations of the Guidelines? How to understand the "reasonable use of generative AI to participate in research and implementation in accordance with regulations" mentioned in the document, and how to make responsible research behavior a conscious action of the scientific and technological community?
It is comprehensive, practical and contemporary
Zhu Bangfen, an academician of the Chinese Academy of Sciences, has publicly stated that responsible scientific research means that scientific research should be responsible to the people and put the long-term interests of the public first. There are many requirements for responsible research conduct, among which research integrity is the most basic.
"With the promulgation of documents on strengthening the construction of scientific research integrity, strengthening the construction of work style and study style, as well as the revision of the Law on Scientific and Technological Progress, the Ministry of Science and Technology, the Ministry of Education and other departments have successively issued documents to build a strategic layout for responsible research behavior. Yuan Junpeng believes that the responsible research behavior mentioned in the "Guidelines" emphasizes the governance of scientific research by the government and society, and also emphasizes the matters that need to be paid attention to in key aspects of scientific research.
The Guidelines put forward scientific ethics and academic research norms that should be followed in carrying out responsible research for scientific researchers, scientific research institutions, colleges and universities, medical and health institutions, and enterprises (hereinafter collectively referred to as scientific research units) from 11 aspects, including research topic selection and implementation, data management, achievement signature, peer review, and supervision and management.
"The drafting and revision of the document took four years, and the name, structure, and content were repeatedly discussed and revised. Yuan Junpeng said that the "Guidelines" originated from the 2009 "Guidelines for the Integrity of Scientific Research Activities", the document was originally named "scientific research integrity", and later found that the content of science and technology ethics, supervision and management, conflicts of interest and other content broke through the scope of scientific research integrity, in order to reflect international practice, the name "responsible research" was adopted.
He Guangxi, a researcher at the Chinese Academy of Science and Technology Development Strategy, believes that the promulgation of the "Guidelines" is an important part of the mainland's efforts to strengthen the construction of study style and work style and the ethical governance of science and technology in recent years.
The Guidelines not only cover the traditional problems of scientific research misconduct such as forgery, tampering, plagiarism, etc., but also include the problems of academic style in the general sense such as impetuousness and exaggeration, as well as the ethical issues of science and technology, that is, the potential risks and adverse effects of science and technology on society. He Guangxi explained that the comprehensiveness of the document is also reflected in the fact that it covers the whole process of scientific and technological activities from scientific research, teaching guidance to scientific and technological applications.
It is worth noting that the Guidelines are not limited to principled guidance, and clearly put forward specific normative standards and prohibited behaviors.
Yuan Junpeng believes that the "Guidelines" fully absorb the academic norms and codes of conduct that have been formed by the mainland science and technology community for a long time and have a broad consensus, and have also been optimized and upgraded for the current scientific research practice. For example, the ethics of science and technology section not only focuses on traditional scientific research ethics, but also proposes corresponding measures for ethical governance in emerging technology fields such as big data and artificial intelligence.
The publication of breakthrough research results must be approved by the institution
For example, for the issue of repeated declaration of scientific research projects, the relevant documents of the state and ministries are not clearly stipulated, and can only rely on the provisions of the project manager. Yuan Junpeng said that the "Guidelines" mentioned at the beginning that the application materials of research projects should be true, accurate and objective, and the same or similar research content should not be used for repeated declarations.
For the publication of high-profile results, the Guidelines make it clear that the publication of breakthrough research results and major research progress should be approved by the scientific research institution where it is located. Research results that have not been scientifically verified or peer-reviewed shall not be disseminated to the public.
This provision has distinctive Chinese characteristics, and it refers to the problem of impetuousness and exaggeration of the style of study. He Guangxi said frankly that in the era of new media, it is a new phenomenon to publish and disseminate the results that have not been peer-reviewed in advance on online platforms such as various preprint platforms. Many researchers believe that the traditional peer review cycle of journals is too long, so they choose to publish their results in advance on preprint platforms.
Although the Guidelines also mention that research results "may be communicated within the scientific community in the form of academic reports, academic seminars, preprints, etc." However, in the process of domestic practice, some media may report some research results that have not been peer-reviewed or even hyped up improperly, which is easy to create a bad impression to the public that the academic style of the scientific and technological community is impetuous and exaggerated. The Guidelines pay attention to this issue and put forward normative requirements. However, the emergence of new forms of scientific research release, such as preprints, also reflects the practical need of the scientific community for faster publication of results. How to better balance the speed of publication and the quality of peer review is an issue that needs to be further seriously considered in future policies. He Guangxi said.
Define generative AI as a research aid
The multiple "roll calls" and focus on generative AI are one of the highlights of the Guidelines. In He Guangxi's view, it is necessary for the document to respond in a timely manner to the problems that may arise from the application of generative artificial intelligence, an emerging technology, in scientific research activities. "Essentially, the Guidelines position generative AI as a research aid rather than a researcher, which is very appropriate. He Guangxi believes that generative AI tools are still a "black box" in the algorithm process, and there is still great uncertainty about the accuracy and authenticity of the process of collecting and sorting data and reasoning to form corresponding results.
In this case, the researcher should be solely responsible for the quality of the results, including the submissions, as generative AI is only a research aid. For example, in traditional scientific research activities, researchers will also have students or research assistants assist in collecting data, but the researcher himself should be responsible for the authenticity and quality of the data.
So, how to understand the "reasonable use of generative AI in accordance with regulations" mentioned in the document? "From the perspective of responsible research, the impact of generative AI on scientific research is still being discussed, and different disciplines also have different views on the rational use of generative AI. Yuan Junpeng believes that the Guidelines have clarified how generative AI should be used reasonably in terms of research implementation, data management, signature and publication of results, and literature citation. As for how to use the scientific research process, it remains to be determined by each discipline according to its own characteristics.
At this stage, Yuan Junpeng suggested that researchers should follow the principle of transparency in the process of using generative AI technology to assist in the writing of papers, and clearly indicate which link and extent the technology is used in the formation and writing of academic results.
On how to speed up the implementation of the "Guidelines", Yuan Junpeng bluntly said that scientific research institutions should earnestly fulfill their main responsibilities, embed scientific research integrity and scientific and technological ethics into the whole process of scientific research, such as formulating specific operating procedures and systems in all aspects of scientific research projects, talent plans, and results publication.
"Researchers also need to recognize that research integrity and ethics are the lifeblood of research, and that flaws in this area can jeopardize careers. Yuan Junpeng reminded.
/ Welfare giveaway, answer questions and receive red envelopes /
A new round of red envelope activities has begun!
Daily 12:00-24:00!
Source: Science and Technology Daily, original title: "Clarifying the "Dos" and "Don'ts" of Scientific Research - Experts Analyze the Guidelines for Responsible Research Codes of Conduct (2023)"
Reporter: Liu Yin
Image Credit: Picture Worm Creative
The cover image and accompanying pictures of this article are from the copyright gallery and are not authorized to be reproduced