1. Introduction
In recent years, massive advances have been made in AI technology-based support for writing research papers. Surprisingly, AIs that can generate drafts of the introduction have been developed, and it is now possible to have the abstract automatically created to some extent [
1,
2,
3,
4]. Could AI-based manuscript writing support undermine the originality and the contribution of researchers? [
5] As to copyright in AI outputs such as texts and images, debate has been ongoing, but the focus has been on the issue of financial rights arising from the creation of copies [
6].
Focusing on AI that offers writing support for scholarly articles, this paper aims to (1) organize representative positions on authorship, (2) examine the potential of AI-based manuscript writing support, and (3) envisage how the role of researchers in future society would change.
There are two major trends regarding the positions of authorship. The first trend reflects that of ICMJE recommendations [
7], and the second, that of
Nature [
8]. There is also a “unique” Japanese position advocated by the Science Council of Japan (SCJ) [
9] (
Table 1). While we certainly think that the first through third ICMJE criteria, as well as the first and second
Nature criteria, should be met, the fourth ICMJE criterion and the third
Nature criterion (i.e., each author is expected to have agreed to be accountable for all aspects of the work, even specific parts of the work in which the author was not personally involved) may be considered somewhat stringent. We predict that it would be difficult to demand perfection regarding this point, and efforts to fulfill these criteria may instead become limiting factors. The SCJ has deemed it appropriate to respect the practices of academic circles, and in terms of author accountability, it only stipulates that “authors are accountable for the contents of the manuscript”. However, detailed provisions are needed, given that the SCJ’s expression is vague and leaves room for interpretation.
From a compliance perspective, journals that conform to ICMJE recommendations require the authors to submit the ICMJE Disclosure Form. Meanwhile,
Nature asks the authors to provide detailed descriptions of their contributions, without specifying the content of contributions to fulfill to be listed as authors. The Consortium Advancing Standards in Research Administration (CASRAI) provides examples of 14 contributor roles (“CRediT”) [
10]. According to
Nature, however, these CRediT roles do not necessarily constitute the criteria of authorship.
Authorship provisions are of interest to all researchers. If the stance of the SCJ deserves some recognition, it is because it expresses respect for the researcher community. To this end, the ICMJE, Nature, and the SCJ will need to listen to the opinions of researchers worldwide, engage in discussion, and strive to increase transparency.
Authorship and originality in research are closely linked. The need to discuss authorship issues arises because each work has original value (“originality”), and it is necessary to specify to whom this originality is ascribed. Authorship belongs to the contribution of a researcher who reports innovative research, and the originality of their work is at the core of their identity. Originality in research is the value of knowledge that the researcher produces.
Writing is considered an important part of work for researchers. Thus, delegating these tasks (e.g., generating a draft of the abstract and introduction) to AI may seem inappropriate, and some may even think this (outsourcing) compromises the researcher’s authorship contribution. The increasing use of AI to this end may eventually generate conflict with researcher creativity and integrity.
But is that true? Does the use of AI when writing manuscripts threaten researcher integrity? If it does, where should the boundary be drawn such that researchers can be warned about their dependence on AI harming their integrity?
2. Discussion
2.1. Originality in the Introduction Section
Does the introduction of a manuscript require originality? Let us assume that it does not and speculate the following: The integral components of the introduction section are the parts that lead to the research question, which require an extensive review of essential research papers, as well as summarizing issues at hand. In this case, it is obvious that creating an introduction with the help of AI support would not compromise researcher originality.
Now let us assume that originality is required for the introduction section of a manuscript. Suppose that the perspective of each author is also reflected in how they perform literature searches and compile search results; in this instance, it may be possible to suggest that the introduction section requires originality. If such is the case, would the use of AI support undermine researcher originality? The answer to this question, in our opinion, is no. Researchers must examine the AI-drafted text sentence by sentence, make necessary modifications, supplement citations, and approve the final version: researcher originality is thus protected.
2.2. AI Support Related to the Discussion Section
The discussion section of manuscripts is where researcher originality is fully demonstrated. Even SciNotes, which offers AI support, restrains itself from writing the discussion section: “Which is why it cannot write the discussion section, which is the most creative and original part of the scientific article and greatly depends on the scientist’s style and way of thinking. Every scientist adds their own expertise and knowledge to the entire text” [
1]. From the perspective of respecting traditional values, receiving AI support in the process of creating the discussion section would mean a deviation from researcher originality.
In what way, then, can AI support be utilized when writing the discussion section? We presume that (1) AI can check for erroneous inferences drawn in the discussion section. Moreover, we wonder if (2) AI will become technically capable of presenting candidates for possible multiple inferences based on results in the future. Of course, we are aware of the technical gap between (1) and (2) (i.e., checking errors in inference vs. presenting candidate inferences).
Next, assume that AI has been created to allow for the writing of the discussion section in the above two capacities. While there is no issue with (1) from the same perspective as that regarding support related to the introduction section, we wonder if AI that offers services such as (2) might undermine researcher originality. Taking a traditional perspective, one may consider that using AI in this way does undermine researcher originality.
However, it is also possible that traditional perspectives do not consider the changing roles of researchers with technological advances in society. We should envision the future, turning our attention to humans who undergo changes in sync with those in technological advances. In other words, it is important to envision the values of future research that can be anticipated at that moment and follow that trajectory in the future society.
Based on the premise that researcher originality is what individual researchers achieve without relying on any environment or tools, we can say that the use of AI undermines researcher originality. However, we must ask whether this premise holds true. In fact, in genome research, studies cannot be carried out without genome analyzers. Furthermore, researchers are influenced by other researchers around them, both explicitly and implicitly, through everyday communication and academic meetings. However, one can imagine that a tremendous number of factors has influenced the researcher’s observations. Perhaps the use of AI is just one of them. This also means that the author’s identity has already been established within the context of environment and society. Thus, the originality of a researcher would no longer be set separately from the environment.
By obtaining AI support, researchers may be able to perform creative work in a more refined fashion. We predict that selecting AI support, evaluating it, and properly adjusting AI would remain an important aspect of work on the part of researchers. Furthermore, even if technology reaches the point where AI can do those tasks, researchers would still need to evaluate its performance. This, which leads to an infinite regress, ensures that researcher originality is protected.
3. Summary
In this paper, we first reviewed the recommendations set forth by the ICMJE, Nature, and the SCJ regarding the way authorship should be credited. In view of these, we tentatively discussed how the development of AI may change the image of researchers. This also affects authorship. The discussion section is the most important part of a manuscript in terms of claiming the originality of the work being reported. Even if advances in technology make it possible for researchers to use AI that offers the discussion section, we argue that this would not necessarily compromise researcher originality.
It resembles caregiving in medical practitioners; even if one day, the era of AI treatment arrives, it is a human wish to be cared for, at least at the end of our lives, by human medical practitioners. This practice (seeing off) would likely never be entrusted to AI even if all other jobs are.
The ethical implication of AI-assisted writing is that developments in AI may renew the researcher’s originality. While AI will advance and improve research, it may also diminish the importance of the author’s role. However, researchers are expected to demonstrate certain originality even when facing such an ethical dilemma. The value of originality has so far tended to be “individualistic”, attributed to independent persons. Perhaps it is time to change this individualistic interpretation of originality and adopt a more “distributed” and “collaborative” meaning. Knowledge production is a collaborative process involving humans, research environments and instruments, and AI. Perhaps the dilemma of AI-assisted writing of papers and diminishing authors’ originality would substantially reform the traditional knowledge framework.