The adjudication of scientific research before artificial intelligence is not the same as after it!
One of the experiences that made me reconsider the academic approach to artificial intelligence was a personal one: I participated in reviewing research papers at the University of Kuwait. With the rapid proliferation of AI tools, it has become difficult to handle research in the same traditional manner, or to assume that everything we read has passed through the familiar channels we were accustomed to years ago. In my view, the question is no longer simply whether a researcher used AI, but rather: How was it used? At which stage? Was it employed to aid in thinking, organization, or editing, or did it replace the researcher entirely? And was this disclosed clearly?
During my time at Princeton University, I observed early on how the emergence of AI sparked genuine debate within the academic community, particularly regarding its relationship with graduate students. The response was not merely about prohibition or accusation; instead, I witnessed sessions, discussions, and brainstorming among faculty and students about what AI could be used for, what the acceptable boundaries are, and what should be disclosed. In my opinion, this is the approach our universities need: transparency before punishment, discussion before accusation, and the establishment of clear rules before holding students or researchers accountable.
Notably, this trend has become even more evident in Princeton’s own policy. The university leaves it to individual instructors to determine what is permitted in their courses, but it requires students to explicitly disclose their use of AI when it is allowed. Moreover, failure to disclose such use can be treated as a violation of academic integrity. The university provides a disclosure template that specifies the tool used and its purpose, such as brainstorming, structuring, or language editing. This shift is significant because it moves the university away from trying to discover “who used ChatGPT?” toward a more useful question: “What did the student do with it, and what part of the intellectual responsibility remains theirs?”
At the University of Oxford, the new research policy goes even further. Since 2025, the AI use policy in research encompasses the entire research cycle, including idea generation, literature review, data analysis, as well as review and evaluation. Oxford requires disclosure of substantial AI use, including the name of the tool and how it influenced the research process, while emphasizing that the researcher remains ultimately responsible for the content. It also warns reviewers against submitting confidential manuscripts or funding applications to AI tools without ensuring the protection of intellectual property and confidentiality.
Similarly, at Harvard, the discussion is no longer about ignoring these tools. University guidelines emphasize the user’s responsibility for all content produced with AI assistance and encourage faculty to clarify what they permit and prohibit for students in academic work.
This leads me to the issue of scientific peer review itself. It may now be necessary for journals, universities, and thesis and research submission forms to include a clear section asking: Was AI used? Which tool? For what purpose? Language proofreading? Data analysis? Literature summarization? Programming? Or drafting parts of the text? Disclosure does not necessarily imply condemnation. On the contrary, disclosure may be the new path to scientific integrity.
AI has become a reality within universities, and it will be difficult to return it outside their walls. Therefore, attempting to ban it completely is not always the most realistic solution. It is more prudent to teach students and researchers how to use it, where to draw the line, how to verify its outputs, and when they should declare its use. Perhaps what we need today is a simple academic rule: Use AI if it is permitted and beneficial, but do not conceal its use, do not let it think for you, and do not claim as your own what you did not produce yourself.
The next phase will not be divided between universities that “use AI” and those that “do not use it,” but rather between those that have succeeded in building a clear, responsible, and transparent culture of AI use, and those that continue to deal with it solely through a logic of prohibition and prosecution.
