Mint finds that 60–70% of firms hiring at five newer IITs now want in-person tests for the 2027 batch, up from 40–50%, because they fear AI-assisted cheating. I read that as backwards. The retreat to geography is a symptom: many large employers in India still lack clearance to use AI in screening at all, so they move the assessment instead of fixing it. I propose the missing layer — randomised, difficulty-calibrated question banks, encrypted district-level test centres, and collusion analytics.
Before booking another train ticket to a remote campus, a recruiter should ask a harder question: has anything been fixed in my screening stack since the last time it was fooled?
Mint reported on 17 November 2026 that placement teams at IIT Jodhpur, Patna, Ropar, Mandi and Indore are seeing a reversal of the online and hybrid model that followed the pandemic. One placement cell member put it plainly:
"Recruiters who previously conducted hiring tests online or followed a hybrid process have asked to visit campuses instead."
The numbers, as Mint gives them:
- Share of firms insisting on in-person tests and interviews for the 2027 batch: 60–70%, against 40–50% a year earlier
- SKF India, Applied Materials and Capgemini were among those indicating offline recruitment
- Day Zero at IIT Bombay, Delhi, Madras, Kanpur, Kharagpur, Roorkee and Guwahati: 1 December 2026
Hindustan Times Telugu syndicated the same story on 18 November 2026. The retreat is travelling faster than any fix.
Distrust is running both ways, too. DCT Japan reported on 17 November 2026 that 22 companies have been barred from IIT placements for two years over withdrawn offers — up from 10–11 last year, with over 150 offers across 23 IITs affected.
Now read the sentence that argues my case better than I can. Kamal Karanth of Xpheno told Mint that many large multinationals in India "do not have internal clearance from their legal and ethics teams to use AI in recruitment." Think about what that means. The return to physical testing is not evidence that physical testing measures better. It is evidence that companies cannot fix their own assessment, and a train ticket is easier to approve than a governance question.
I can agree with Atul Sahgal of Cognizant that "a resume or virtual interaction is not a substitute for a conversation across the table" — and still refuse the binary. That sentence defends the last hundred metres of hiring. It does not defend carrying the first two screening rounds across four states.
I have been building the other answer for years. In November 2012 I was describing interview consoles that recorded a recruiter's rating and mailed the offer the same minute — and, even then, about eliminating "proxy interviewees". The National Employment Grid whitepaper of July 2025 turned that into InterviewBots that conduct first rounds and return percentile scores. ResumesExchange, drafted in April 2026, makes the resume pool a shared index instead of a rented one. What is still missing is the assessment layer itself.
So, instead of a travel layer, I would suggest:
- Per-candidate randomised question sets from a difficulty-calibrated bank, generated seconds before the test — the Central Exam Conduct Board idea in Cheating in Exams : is only a Symptom (March 2024).
- Encrypted delivery to a district-level standardised testing centre, the school-network model I proposed for the NTA in October 2024, so a candidate near Mandi or Patna sits an invigilated test without a day-long journey.
- Post-hoc analytics on identical-answer clusters — the same wrong option chosen by strangers, matching timing curves.
Do this and no employer has to choose between online and trustworthy. A better assessment will always beat a longer journey.
With regards,
Hemen Parekh
www.Iam-Immortal.ai / www.IndiaAGI.ai / www.YourContentCreator.in
18 November 2026
No comments:
Post a Comment