Exploring AIGC Academic Ethics: Triple Constraints of Government, Publishing, and Universities

On September 7, 2023, UNESCO released the “Guidelines for the Application of Generative Artificial Intelligence in Education and Research“^[1], which details the definition of generative artificial intelligence, discusses the controversies it raises, and its impact on education, particularly the issue of exacerbating the digital divide. The guidelines propose key steps that governments should take to regulate generative artificial intelligence and suggest establishing a policy framework to ensure that generative artificial intelligence is applied ethically in education and research. The guidelines also recommend setting the minimum age for using AI tools at 13 years old and call for relevant training for teachers.

The Ministry of Science and Technology’s Supervision Division and the National Natural Science Foundation of China also released the “Guidelines for Responsible Research Conduct (2023)“^[2] and “Research Integrity Standards Handbook“^[3] in December 2023, clarifying the norms for using AIGC, including research topics, data management, and authorship, and emphasizing the need to use AI tools cautiously during evaluation activities. An excerpt from the “Guidelines for Responsible Research Conduct (2023)” is as follows:

On September 20, 2023, the China Institute of Scientific and Technical Information, in collaboration with Elsevier, Springer Nature, and Wiley, released the “Guidelines on the Boundaries of AIGC Use in Academic Publishing“^[4], which clarifies the boundaries of AIGC use in academic publishing and aims to guide consensus on the norms of AI technology usage among the publishing industry, scientific community, and science and technology management departments. This guideline emphasizes that although AIGC can assist in data collection, statistical analysis, chart creation, and text editing, researchers must verify and edit the outputs provided by AIGC to ensure originality and accuracy and must disclose this appropriately, and cannot be listed as authors of papers.

Major publishers (Cambridge University Press^[5], Elsevier^[6], Emerald^[7], IEEE^[8], Sage^[9], Science^[10], Springer Nature^[11], Taylor & Francis^[12], Wiley^[13], etc.) follow the guidance of the Committee on Publication Ethics (COPE)^[14], which does not allow AIGC tools to be listed as authors of papers because AI cannot assume legal and authorial responsibility. Publishers require authors to disclose the use of AIGC in the methods section or acknowledgments of the paper and take responsibility for the objectivity and accuracy of AIGC-generated content.

Although the publishers’ policies are strict, they also provide room for the reasonable use of AIGC. For example, Elsevier, Emerald, IEEE, and Wiley state that AI can be used as a proofreading tool to help improve the linguistic quality of articles but cannot replace the authors’ critical work.

Furthermore, the detailed regulations vary among publishers; for instance, Elsevier specifies that AI tools cannot be used to modify or introduce features of images but can adjust brightness or contrast as long as the original image information is not affected; whereas Emerald, Science, Springer Nature, and others explicitly state: AI tools cannot be used to modify or generate multimedia materials.

Therefore, researchers must carefully read and comply with the relevant regulations of publishers before publishing papers to ensure that academic publication is not adversely affected by neglecting these rules.

Some universities abroad initially opposed the use of AIGC in the classroom, but over time and with the widespread use of AI tools, their attitudes have shifted to supporting the use of AIGC under certain conditions (such as proper citation, data protection, avoiding AI hallucinations, and compliance with academic integrity policies of the institution or instructors).

Notable institutions such as Harvard University^[15], University of California, Los Angeles^[16], Princeton University^[17], Boston University^[18], the Group of Eight in Australia^[19], and universities like Oxford and Cambridge^[20] have established guidelines to guide faculty and students in the correct use of AIGC.

Harvard University has released guidelines for the use of AIGC tools.

In China, universities such as Shanghai Jiao Tong University^[21] and ShanghaiTech University^[22] have published guidelines to provide clear guidance for faculty and students.

To maintain the originality, fairness, and legality of academic research, governments, publishers, and educational institutions collaboratively construct a multi-dimensional framework. As AIGC (Artificial Intelligence Generated Content) technology continues to advance, corresponding norms and guidelines are constantly updated to adapt to new developments. As members of the academic community, we have the responsibility to continuously learn, understand, and adhere to these norms. Only by doing so can we fully utilize the conveniences and innovations brought by AI while jointly promoting the healthy development of academic publishing and ensuring respect for academic integrity and ethical responsibilities.

This is a preliminary exploration of AIGC academic ethics, and I look forward to further exchanges with everyone. Please feel free to leave comments, share your insights and suggestions.