In India, AI is only exacerbating academic inequality

One interesting argument to emerge from the many deliberations on artificial intelligence that have been published in the past few years is that the technology will help people who have historically struggled with the English language.

That is a particularly pertinent issue in India, where since the colonial era English has been a marker of status and merit. There is a constant policing of tone, accent and grammar, and academic positions in the social sciences are often determined by the way one speaks and writes in English. Those who lose out in these circumstances are usually the most marginalised: women, Dalits, Adivasis, tribes and Muslims.

Yet while AI may appear to be democratising academic access, my experiences of teaching at a leading technological public university, as well as being on the editorial board of a public sociology blog and an academic journal, suggest that the reality is less positive. Observations from the classroom, PhD interviews, reading assignments, PhD proposals, blog articles and journal articles underline that the effective use of AI still depends on forms of knowledge that are themselves unequally available. Everyone has instant access to AI, but not everyone has informed access to it.

The unequally distributed digital academic capital includes knowing how to verify sources, understanding academic ethics and distinguishing between editing and collaboration. It also includes knowing how to use AI without getting caught. Those unfamiliar with the digital world don’t know, for instance, that the em dash is considered to be a marker of AI, or even that AI is prone to fabricate references. I have had conversations with students who confessed that they were not aware that AI could ever be wrong.

In reality, of course, learning to compile reference lists is itself an academic skill that everyone should learn, rather than outsource to AI. But those who have not had rigorous methodological and academic training do not know how to do so, and their naive equation of what AI generates with the truth gets them into disproportionate amounts of trouble.

For instance, India’s University Grants Commission (UGC) recently issued revised guidelines to regulate the use of AI and guard against plagiarism in PhD theses. It clarified that the main content of theses must be written by the researcher, with AI permitted only for language correction. Violation of these norms could result in the cancellation of the scholar’s registration and the suspension of their supervisor’s right to supervise PhDs. This is a welcome step, but we need to be cautious before celebrating its implementation.

My experience suggests that awareness of these rules will never become widespread. Even knowledge of basic academic standards is often lacking. For instance, I once received a submission to the journal that I was editing consisting of only two paragraphs of bullet points. When I rejected it, its author wrote to the director of my institute and to the higher authorities at Springer seeking an explanation of why the article had been rejected.

This situation is rooted in the Indian higher education system. Training is rarely offered on research standards or ethics, let alone AI use, especially in regional education institutions. So would it really be fair to punish researchers for using AI in their theses for more than language correction – especially those from the margins? This is an especially pertinent question in a country where the PhD is primarily considered to be a degree to obtain a job, rather than academic training.

It is also important to bear in mind that PhD places are few and far between in India – as is guidance on developing research proposals. As such, researchers use AI uncritically and unethically as a practical tool to navigate a highly competitive selection process, without fully understanding the risks of fabricated references or breaches of academic ethics.

Moreover, even when students and researchers understand academic right from wrong, the publish-or-perish culture that exists in most academic institutions still pushes them towards unethical practices. Many faculty, and especially early career scholars and those on short-term contracts, are burdened with teaching, managerial and administrative work. At the same time, they are expected to publish to further their careers. This means that they often resort to plagiarism, ghostwriting and predatory journals. Now AI makes bulking up their publication lists even easier – and the promised rewards seem to outweigh any concerns about academic ethics.

That is especially true for those without academic cultural capital, who don’t understand that once you are caught publishing plagiarised or AI-generated text in a predatory journal, your reputation is tarnished and your future research and job prospects suffer.

Thus, while AI may be reducing linguistic inequality, the benefits of that may well be outweighed by the additional ethical tripwires it sets up in the paths that the underprivileged must walk towards academic success in India.

Rituparna Patgiri is an assistant professor in sociology at the Indian Institute of Technology Guwahati.