Indian Academia’s AI Surveillance System is a One-Way Mirror. Here's How

Last year, a student displayed ChatGPT proudly on the screen at his graduation ceremony at the University of California, Los Angeles, and faced a lot of backlash on account of the increasing use of artificial intelligence (AI) by university students and the devaluing of education. 

In the beginning of September this year, New York City mayor Zohran Kwame Mamdani and schools chancellor Kamar H. Samuels announced a ban on the use of generative AI by students up till high school, with usage even in high school being limited. However, teachers are allowed to use it for instructional planning and operational tasks.

Last year, I had asked a professor for a letter of recommendation (LOR) for a competitive fellowship. When I received the letter, it registered as AI-generated: the sweeping generalisations, the hedged transitions, the enthusiastic but oddly impersonal endorsement, all sounding more like a probabilistic summary of my resume fed into a prompt. To this, the reply I received was that AI could do a better job than me, leading me to think about AI-use policies and their consequences. 

This irony of AI-use for professors and students is very structural. A student is subjected to thick AI-disclosure requirements and academic misconduct policies, be it an essay, a dissertation or exam answers. As a student, I have voluntarily made many disclosures on various platforms before submitting my work, and yet the materials flowing from the other direction: lecture notes, syllabi, course readings, doubt clearances and now letters of recommendations are not subjected to such accountability.

This surveillance, which seeks to ensure that academic integrity is maintained, is rather a one-way mirror that puts faculty in a blind spot. 

Mapping the policy asymmetry

The recent news of death by suicide of a second-year student at IIT-Bombay also raises many questions about the drastic consequences of AI-use for a student. This seemingly symmetric acceptance of AI-use is worth looking into; it comes winding down only on students and their outputs, while professors, teachers and evaluators do not go through similar screenings. 

The foundational document governing academic integrity at higher educational institutions in India is the University Grants Commission's (Prohibition of Academic Integrity and Prevention of Plagiarism in Higher Educational Institutions) Regulations, 2018. Its preamble states that the regulations apply to students, faculty, researchers and staff alike, however the operational weight is much different. 

The penalty structure is also elaborate, from 'level 1' requiring revisions, to 'level 3' resulting in the total cancellation of a student’s registration. For the faculty, existing penalties only concern their research publications, including debarment from supervision, for instance. But even then, there is a conspicuous absence of any mechanism governing the daily academic labour of a professor: the syllabi, the lecture slides, the classroom materials and the documents they produce on behalf of the students. There are no regulations mentioning an AI-generated LOR, despite the document carrying major credential weight for a candidate’s future. 

There are institutional policies governing AI usage, where this gap is unusually clear. The AI policy at IMT Nagpur provides the starkest comparison in this regard. Under the student-facing provisions, a detailed list of obligations reveal which tools were employed and when. Under the faculty-facing provisions, on the other hand, there are recommended practices, suggestions for designing better assessments and an overall encouragement to educate students about 'responsible AI-use.' Similar guidelines are present in the AI-use policy issued by the Indian Institute of Bombay, specifically saying: 

When there is any case of inappropriate AI use by students/research scholars, instructors/supervisors must report the academic malpractice…Disciplinary measures will be decided as per the gravity of the malpractice and as per prevalent institute rules and regulations. 

Several AI policies require the faculty to learn more about using AI to mark students’ papers, the extent and specifics of which are unknown as AI is known to be biased. 

Therefore, a student must declare what specific tool they used and how it was employed, while a professor is absolved of any such declarations pertaining to lecture notes, course materials and LORs. This silence is hardly an oversight; it is yet another instance of how higher education institutions in India are still heavily biased against students. 

The blind spot

This institutional asymmetry is mirrored in the research literature. There are countless existing reports and articles on the increasing use of AI in higher education institutions around the world, talking about the end of critical thinking in classrooms, and almost all of them are overwhelmingly focused on students. Data on faculty AI use is essentially absent, especially in India.

Thus, the query of rising AI use and its effect on educational integrity and intelligence has not yet been posed in a symmetrical, all-encompassing way. 

International research presents its own skewed views. A survey by the EdWeek Research Centre found that 80% of educators did not believe they needed to disclose AI use in preparing lesson plans, to either students or parents. Many also described AI as an efficiency tool, but this framing warrants a careful examination as despite claims of AI efficiency, it is essentially not a reasonable tool. 

A recent LLM graduate from a tier-1 law school in India, recalled: "I was in a seminar course. The professor openly kept ChatGPT open on his laptop the whole time, and whenever anyone asked hime a question, he fed it into the AI and gave us an answer. This was such a waste of our time."

When such incidents happen, where are the queries surrounding educational integrity and cognitive impairment due to AI? 

The credentialing integrity crisis

Though students have to face severe consequences for any AI use, they cannot question the opacity of grading parameters. They must not submit AI-generated answers, but the questions themselves can be AI-generated. Looking at the different forms of asymmetry prevalent in AI-use, it is useful to analyse the consequences of an  AI-generated letter of recommendation . 

There are several scholarships, fellowships and admissions applications that strictly say that an application will be disqualified in case AI-use is detected. Then, if the candidate is careful enough not to do that and the letter of recommendation is still AI-generated, who bears the consequences? 

AI is used to grade papers, give comments on an assignment, filter applications based on a prompt and then used to pass off a probabilistic summary generated by a machine as a 'professional personal assessment.' The word “fraud” would not be out of place here. The question I posed in the previous paragraph is then not abstract: it directly related to the student who loses. 

When the lecture notes carry the characteristic cadence of a large language model, they signal something about the intellectual labour that into them. These are the materials that we as students build our understanding of the course upon and they themselves end up being a machine-written.

If academic rigour was taken seriously, this same process, like for students, would have resulted in a disciplinary committee and dire consequences. 

One-sided integrity crisis

Why is it that an AI-generated assignment gets more wrath than an AI-generated lecture? The intellectual honesty demanded of a student cannot exist in this kind of one-sided integrity.

If a student’s essay requires an AI-statement, then a teacher must be required to disclose the usage of AI as an evaluator. If one must be debarred from an application for using AI, then the one who produced an AI-generated letter of recommendation must face consequences as well.

Either academic integrity is a shared value that structures student and faculty conduct in the institution symmetrically, or it is simply a mechanism of control dressed as a rule. Both positions cannot be maintained simultaneously. 

Nishita Sharma is a recent graduate from NALSAR University of Law and is currently working as a law researcher. 

The Wire is now on WhatsApp. Follow our channel for sharp analysis and opinions on the latest developments.