The interactions of AI and acdemic research facilitate academic exploration. Photo: TUCHONG

---

The impact of artificial intelligence (AI) represented by ChatGPT has permeated people’s lives, and as a result, comprehensive reflection on it has been initiated in various academic fields. Whether actively embraced or passively influenced, the development trend of AI is clearly laid out before humanity, and it is unstoppable.

ChatGPT as a tool  

Although OpenAI’s description of ChatGPT highlights its creative writing capabilities (“it can generate, edit, and iterate with users on creative and technical writing tasks, such as composing songs, writing screenplays, or learning a user’s writing style”). In theory, the essence of listed tasks is to write according to a predetermined pattern or framework. Does this represent creative writing? Yes, but its creativity is reflected in the homogeneous output, rather than in exploring the unknown, and is fundamentally not detached from imitation.

Therefore, as far as the learning algorithms currently used by ChatGPT are concerned, it has still not transcended its essential nature as a technical tool, and its output does not constitute academic research.

AI validity for academic research

The use of AI tools like ChatGPT does not constitute knowledge innovation. Compared to other industries, the academic community is more sensitive to ChatGPT, with acceptance and resistance coexisting. A notable manifestation of this is that the world’s most important academic journals prohibit ChatGPT from being listed as a collaborator and oppose the use of ChatGPT-generated papers. However, some advocates view ChatGPT’s conversational text generation as a form of knowledge production.

These two perspectives are not actually contradictory, as they address different levels of content. The academic journals that prohibit ChatGPT from being listed as a co-author of papers and oppose the use of its generated texts represent the latest exploration of the mysteries of the universe and humanities, the discovery and verification of the unknown in the natural and humanistic fields, and the field of knowledge innovation. At this level, the opposition and prohibition of ChatGPT as a co-author of papers and text generation reflect a reverence for knowledge innovation.

The advocacy of ChatGPT’s conversational text generation as knowledge production highlights the unlimited value of AI as a tool for skilled work. It should be pointed out that this type of knowledge production involves repetition and transfer within its own category, rather than creating the unknown. Therefore, strictly speaking, the text generated by ChatGPT does not constitute true knowledge production, but rather the application of knowledge.

Based on the current foundation and operational effectiveness of the ChatGPT algorithm, even as it becomes increasingly sophisticated, its validity and limitations in academic research can be roughly anticipated.

Firstly, in terms of language expression, ChatGPT can function as a relatively stable and valuable tool for academic research paper writing.

Secondly, the instrumental value of ChatGPT in academic research varies greatly. In disciplines that use formatted language as a tool, it holds the greatest value. As for social sciences that use data collection and statistical analysis as tools, ChatGPT can also effectively leverage its instrumental value. In terms of humanities, ChatGPT can serve as a tool in some fields, while in others, even if the input data is comprehensive enough, current algorithms not only fail to achieve a useful level of creativity, but even the functionality of the tool is difficult to fully realize.  

Finally, with a comprehensive understanding of contemporary scientific development and academic research as a whole, it becomes clearer that the advancement of AI is an upgraded version of the 20th-century “quantitative revolution.”

Appropriate utilization

The development of AI, represented by ChatGPT, is a product of the marriage of supercomputing technologies and an algorithm revolution, essentially constituting a new phase of the “quantitative revolution.”

In addition to the aforementioned positive aspects, ChatGPT naturally presents negative implications for current academic research. However, it should be understood that technologies or tools themselves are neutral, and their value lies in the academic cultivation and ethics of tool users, as well as the recognition and judgment of academic innovation by the academic community.

From a holistic perspective of human development throughout history, each advancement in measurement and algorithms and every progression in technological tools have further propelled the exploration of the mysteries of nature and humanity. From the standpoint of a nation or a country, those who embrace sci-tech innovation often experience flourishing development.

Therefore, although current AI represented by ChatGPT is still far from being perfect and even subject to many jokes in academic research—such as generating the academic research history of a specific field, where the listed publications are often not representative academic achievements in that field—it should be appropriately utilized and improved in academic research as a trend of scientific revolution and technological advancement, rather than being rejected.

Although I am a conservative humanist, my academic training allows me to understand that AI, no matter how powerful, is only a technical tool for human use. While the use of tools itself is not innovative for academic research, it can help facilitate innovation in academic research by significantly improving the methods and means of human exploration of nature and human mysteries.

Pan Sheng is a professor from the School of Social Development at Nanjing Normal University.

Edited by ZHAO YUAN