An IIT Madras white paper says AI will not take jobs, only tasks — but the tasks it takes first are the ones a junior learns judgement on. Reading it beside TCS and Infosys fresher numbers and Nandan Nilekani's remark, I make three suggestions: a structured judgement apprenticeship with an intake floor, a paid "AI supervisor" cadre, and annual task-level disclosure by employers, so we can see the ladder being shortened before the middle goes missing.
The Times of India report on a new IIT Madras white paper quotes B Ravindran, who heads the Centre for Responsible AI, saying something I have waited years to read from an Indian institution: "AI will not take away jobs. Most jobs have tasks, and only some of those tasks can be undertaken by AI."
Then he says the thing that worries me far more: "Entry-level jobs will shrink if we do not reimagine how we work." The paper's own executive summary admits "the entry-level career pipeline is especially at risk". The tasks AI absorbs first — password resets, OTP checks, invoice lines, test scripts — are precisely the ones on which a young worker used to build domain knowledge before being trusted with judgement. As the TOI report of 26 October puts it, those routine assignments "often help new workers acquire domain knowledge before moving into more complex roles".
I have argued the whole-job version of this for years — see Wherefore Art Thou, O Jobs? (May 2023) and Graduates Without Jobs (March 2026). Both posts argued about the destination. This paper argues about the staircase, which is the better question.
The staircase is already being pulled up. According to CNA, reporting on 27 October, TCS hired more than 110,000 engineering graduates four years ago, and its FY2026 intake was less than half that level; Infosys took in about 80 per cent fewer freshers in FY2025 than three years earlier. At the Bloomberg New Economy Forum on 13 October, Nandan Nilekani said AI was increasingly taking over the coding juniors once did.
Then Sumit Sarawgi of Oliver Wyman asked the question that belongs on every board agenda: "Where will the future middle managers come from?" In his firm's survey, 45 per cent of Indian managers said AI was taking over tasks that once built junior experience, and only 12 per cent of respondents said their employer had a plan for the time AI frees up. Forty-eight per cent had been given no guidance at all.
Three things I would suggest, all of them cheap against the cost of an empty middle:
| Proposal | Who acts | What changes |
|---|---|---|
| Judgement apprenticeship | Employers, with sector councils | A structured two-year first job — rotation, a named mentor, defined sign-off authority — with an intake floor even when AI does the routine work |
| AI supervisor cadre | Employers and skilling bodies | A paid, graded role for people who verify and escalate AI output, so verification becomes a career step instead of a dead end |
| Task-level disclosure | Employers, annually | Publish which tasks were automated, which were redesignated and which were kept deliberately for training |
I am not asking firms to protect tasks that AI does better. I am asking them to protect the learning those tasks happened to carry, because that is a design choice and not a technology outcome. CeRAI's Gokul Krishnan is right: the outcome depends on capability, cost, safety, regulation and "how organisations restructure operations". The restructuring is the part we control.
In 2017 I quoted a TeamLease study in which employers rated apprentices more employable than freshers (Discovering the Obvious?). If that was true when jobs were plentiful, it is urgent now. Sarawgi's question does have an answer — future middle managers will come from the juniors somebody deliberately decided to keep.
With regards,
Hemen Parekh
www.Iam-Immortal.ai / www.IndiaAGI.ai / www.YourContentCreator.in
27 Oct 2026
No comments:
Post a Comment