Sudipto Mundle: How artificial intelligence has begun to reshape India’s challenge of job generation

Around this time last year, in my presidential address to the Indian Econometric Society [Journal of Quantitative Economics, 2025, 23:319–331 bit.ly/4s4McEJ], I had discussed India’s growth paradox: namely, that though India has been the fastest growing major economy in the world for quite some time, the number of unemployed in the country has also been growing almost as fast.

To address this employment challenge, I had outlined a strategy based on three pillars. Within just a year, I now need to revisit that strategy, not because the pillars have changed but because the content of each pillar has to be adapted to a rapidly changing employment context driven by the rising tide of AI.

The first pillar I had outlined, the short-term one, was the need to deploy ‘industrial policy’ for the non-agricultural sector to help accelerate the growth of employment-intensive industries in addition to high-tech industries and services. The latter add a lot to GDP but relatively little to employment.

In comparison, just six employment- intensive sectors account for as much as two-thirds of all employment outside agriculture: construction, trade, land transportation, processing of food and beverages, apparel manufacturing, hotels and restaurants. Most jobs in these sectors require a low level of skills. The corresponding level of productivity and remuneration is also low, but their low skill requirement matches the skill profile of the bulk of India’s labour force, which cannot be changed overnight.

Accelerating the growth of these sectors will quickly expand the scale of employment and livelihoods, however modestly, in the short run; this will buy India time to appropriately skill the labour force and prepare it for higher productivity and better-paid jobs in the medium to long term.

Given the emerging labour market impact of AI, it is possible that these employment-intensive sectors will grow faster even without the prop of ‘industrial policy.’

At the recent AI Summit in New Delhi, Kristalina Georgieva, managing director of the International Monetary Fund (IMF), made some very important observations.

The IMF, she noted, is finding that AI will affect—not necessarily eliminate—40% of jobs globally. More interesting is its finding that demand for very high-skill jobs and that for low-skill jobs is actually increasing with the impact of AI, while demand for entry-level skilled jobs in the middle is disappearing. Human labour is being displaced by AI in these tasks.

Why demand should be rising for mass consumption products when income inequality is increasing is not clear. But if true, then in the Indian context, this would imply that demand for goods and services delivered by the employment-intensive sectors cited above would accelerate. That’s the good news.

The bad news is that, by the same token, the second pillar of my suggested employment strategy will be much more challenging. I had proposed a skilling strategy called University Technical Education (UTC).

This is an alternative to the conventional path of secondary education which could replace the current dysfunctional vocational education programme. UTC would be a private sector-led, self-financing programme. It would combine STEM-oriented classroom education with intensive shopfloor training in operational plants of corporate partners. Students graduating from this alternative higher secondary education programme could go on to higher education if they so desired or opt for employment.

In the latter case, their UTC graduation combined with shopfloor training would better position them for jobs than conventional higher secondary graduates. The UTC programme, along with a significantly reformed higher education system, could prepare India’s workforce for the 21st century.

The broad structure of this second pillar would remain intact despite AI. But the content of the STEM-oriented courses would have to prepare students not for entry-level jobs but higher skills for use in roles where they would have to prompt AI models to perform elementary tasks. The same would apply to on-the-job training on the shopfloor.

But AI models are evolving so rapidly that today’s training in their use could become obsolete tomorrow. The UTC courses would therefore need to be designed in collaboration with experts from AI companies who can see in which direction AI models are headed.

Fortunately for us, the fact that so many corporate leaders of Indian origin are embedded in leading global tech companies places India exceptionally well to move down this path. The development of India’s own foundational models like Sarvam is excellent news and should help the country prepare its workforce even better for the future.

My third pillar was raising productivity in agriculture. No matter how the employment outlook outside farming might improve, agriculture will continue to be a major sector of employment for years to come. However, productivity is abysmally low in agriculture.

I had earlier discussed standard approaches to ramping up productivity. However, application of AI could radically transform traditional farming—from water and land management to selection of seeds and crops, production techniques, post-harvest technologies in storage and transportation, etc.

This AI revolution in agriculture, not unlike the ongoing revolutions in biotech, medicine and other sciences, could raise labour productivity in ways unimagined so far.

These are the author’s personal views.

The author is chairman, Centre for Development Studies.