India’s AI transition needs paid-practice floor for young workers

India’s debate on artificial intelligence (AI) often swings between two aggregate stories: mass displacement and mass opportunity. Recent reporting shows why neither is enough. Mint’s coverage of AI-related hiring and layoffs suggests India is creating significant AI-linked roles while routine support work is being automated and demand shifts towards people with more context and experience.

The Stanford Digital Economy Lab’s August update sharpens this problem. By June 2026, employment for workers aged 22 to 25 in highly AI-exposed occupations was about 19% below the path implied by less-exposed occupations, compared with 15% in the July 2025 data vintage. The adjustment appeared mainly through reduced hiring rather than separations. That is not an economy-wide job-loss estimate. It warns what can happen to the first rung of selected career ladders when routine work is automated faster than new learning work is designed.

For a country with millions of young labour-market entrants, access to practice is therefore a distributional issue. A worker who completes an AI course but cannot obtain the first paid assignment has not received the same opportunity as someone who learned the old workflow before automating it. The second worker owns tacit knowledge, references and examples of judgement. The first may have a certificate but no evidence that an employer should trust them with a difficult case.

A paid-practice floor would attach a simple condition to public support for AI adoption. If an employer benefits from subsidised training, digitalisation incentives or publicly backed AI infrastructure in a highly exposed function, a portion of the programme should fund supervised early-career assignments. These would be real jobs or paid projects, not unpaid simulations.

The assignments should preserve the parts of work that create leverage for workers: handling exceptions, talking with customers or citizens, validating source data, documenting errors and learning when a model’s answer cannot be accepted. AI can handle routine preparation, but the worker must own a consequential piece of the final process under a named supervisor.

Government can evaluate the floor using labour-market outcomes rather than training volume alone: conversion to paid employment, wage progression, time to independent responsibility, retention and the percentage of AI-exposed teams still hiring at entry level. If productivity programmes accelerate while these measures collapse, public policy would be subsidising a short-term efficiency gain and a long-term experience shortage.

Technology policy should not romanticise repetitive work. It should recognise that paid practice has economic value because it is how workers acquire the judgement that markets later reward. India can automate routine tasks aggressively while still guaranteeing that young people get a paid route into the experience economy that AI makes more valuable.

The floor could be financed partly from productivity gains. When a public programme helps an employer automate a process, the business case already estimates time or cost saved. Setting aside a small, declining share for paid practice makes the social investment self-limiting: as workers become independently productive, the subsidy falls, and the employer captures more of the gain.

Worker representation should be part of programme design. Employees know which routine tasks are pure drudgery and which are where beginners learn the trade. That distinction cannot be inferred reliably from a task label. Consulting workers before redesign can therefore improve both efficiency and training by automating the least developmental work first.

The floor can also recognise informal-sector transitions. Many young Indians build experience through small enterprises, freelancing or family businesses before entering larger firms. Approved projects should include those settings when a credible supervisor and real customer outcome exist, so the policy does not privilege only formal corporate pathways.