UGC NET cancelled papers written by AI? Experts flag signs of AI use, NTA denies

The National Testing Agency has cancelled and ordered fresh UGC-NET papers in English, commerce and sociology after candidates and subject experts have flagged a string of errors, including badly mangled names of well-known sociologists and dozens of repeated questions lifted from earlier exams, according to sources cited by TOI. Insiders say the papers were prepared with heavy, poorly supervised use of artificial intelligence, though NTA has denied that generative AI was used to author them.

UGC-NET exam decides who is eligible to teach and pursue research at universities across the country, which is why a mix-up of this scale has turned into more than just an embarrassing footnote. It has become a question about how much of India's biggest exams are now being built by machines, and how little human checking may have happened afterward.

When Ritzer Became "Putzer"

The sociology paper drew the sharpest reaction. Candidates spotted names of major scholars twisted almost beyond recognition, George Ritzer turned up as "Putzer," Talcott Parsons became "Parsow," and G S Ghurye was rendered "Ghunye." Martha Nussbaum's name did not survive the process either. Beyond the spelling, students said the terminology felt garbled and the questions themselves barely matched what they had actually studied. It reads almost like a comedy of errors, except the people affected are researchers and future academics whose careers hinge on this one test.

The Case of the Repeated Questions

The English paper had a different problem: repetition, and a lot of it. Candidates alleged that around 67 of its 150 questions had shown up in an earlier UGC-NET exam, with some claiming even the order of the answer choices matched. Commerce candidates reported something similar, pointing to repeated questions on taxation, break-even analysis, capital structure and services marketing. Nobody has yet explained exactly how the same questions ended up back on the paper.

Ministry Puzzled With Six Weeks of Silence

What has puzzled the education ministry as much as the errors themselves is the gap before anything was done about them. Ministry officials reportedly wanted the flawed papers cancelled and retests scheduled by the end of July or the first week of August. NTA did not act on that advice. It waited nearly six weeks, and only announced the cancellations when it released provisional answer keys for 84 of the 87 subjects on offer.

A senior source in the education ministry described meetings with two senior NTA officers who admitted they could not finalise the answer keys because the papers had been "designed in an ad hoc manner." Pressed on what that actually meant, the source said, the officers had nothing further to offer.

Another source raised a more basic question about how the papers were even assembled in the first place, asking pointedly "how did the examiners not have answers to the questions they had designed?" A panel of four or five experts is supposed to set each paper, so if that panel existed, it is unclear how so many recycled questions slipped through unnoticed.

NTA Says No AI, But Questions Remain

NTA director general Abhishek Singh has rejected the idea that generative AI wrote the disputed papers. He told TOI there was a "human review" once the papers were set, but when asked how the errors then went unnoticed, he did not elaborate further.

A faculty member from JNU summed up the deeper issue, noting the real question is "whether adequate human academic review took place after that process was complete." Reliance on technology, on its own, does not prove AI caused the mistakes.