AI could wipe out demand for labour in 5-10 years, warns economist Kaushik Basu

Artificial intelligence could dramatically reduce the demand for human labour within the next five to 10 years, creating unprecedented levels of inequality unless governments rethink how the wealth generated by machines is distributed, economist Kaushik Basu warned at Godrej Literature Live in Mumbai.

In the last session of this season, in a conversation with journalist Govindraj Ethiraj at the Mumbai literature festival, Basu, a former chief economist of the World Bank and former chief economic adviser to the Government of India, said the world could be heading towards a future in which factories operate almost entirely through artificial intelligence and robotics, generating enormous wealth for their owners while leaving much of the population without jobs or purchasing power.

“The demand for labour ... it is frightening; it is frightening,” Basu said, cautioning against the assumption that technological revolutions would automatically create new employment opportunities, as they had in the past.

His argument was that India must pursue two strategies simultaneously: exploit the immediate opportunity to expand manufacturing and generate employment, and begin preparing for a longer-term economic order in which the relationship between work, income and wealth ownership could change fundamentally.

Basu argued that India still has a window of opportunity to expand its manufacturing sector as global production networks shift and China vacates some of the space it once occupied.

Pointing to Vietnam as an example of a country stepping into this space, he said India could similarly attract manufacturing activity and generate substantial employment if it addressed the practical barriers to doing business.

“We are seeing one great example, Vietnam is stepping into the space that China is vacating. India can get in there,” he said.

This would require a closer examination of labour laws, the regulations governing the establishment and operation of businesses, and other micro-economic factors that affect manufacturing competitiveness, he said.

India has made progress on some fronts, but more detailed reforms are needed to capitalise on the opportunity.

For Basu, manufacturing is not merely an industrial policy priority but a way to buy time before the full employment implications of AI become apparent.

“For the next five years, we should exploit it. Put our workers there, build factories, there's scope,” he said.

He argued that the expansion of manufacturing could create demand not only for workers on factory floors but also for managers and others involved in running production systems.

However, he cautioned that the opportunity would not last indefinitely. As AI systems become more capable, India, like other economies, will have to confront the possibility of a much sharper decline in demand for traditional labour.

Basu's concern is that the economic gains from AI may accrue disproportionately to those who own the technology, intellectual property and productive assets.

Citing inequality research that tracks the wealth of the world's richest one per cent, he pointed to the growing concentration of wealth among them, and argued that the problem could become far more severe if businesses could produce goods and services with little or no human labour.

“If you look at them in the past, the demand for labour vanishes,” he said, describing a possible future in which factories producing steel and automobiles operate with machines running the entire process.

In such a scenario, the world could become extraordinarily wealthy in aggregate, even as large sections of its population lose their principal source of income.

“Who’s getting all the wealth? It’s people who own the shares. It’s their companies doing it. People who own the intellectual patents and property rights, they are getting it all,” Basu said.

The result, he warned, could be a level of inequality in which a tiny group controls an overwhelming share of economic resources while the majority struggles to participate in the economy.

Basu argued that addressing this would require moving beyond conventional welfare measures towards a system that redistributes a portion of the profits generated by increasingly automated economies.

“The only way to correct this is to take away a part of the shares from the super rich (and distribute it) across the population,” he said.

He clarified that he was not advocating the abolition of private enterprise or a system that eliminates incentives to innovate. Instead, he called for an intelligent system of taxation and redistribution that could spread the gains from technological progress without destroying the motivation to build businesses.

He also stressed that such a system could not work effectively if implemented by individual countries in isolation, as capital could move to jurisdictions with more favourable tax arrangements.

“We need to bring the whole world together,” he said, arguing that countries would have to negotiate common rules to prevent capital flight and make the redistribution of AI-generated wealth feasible.

Basu suggested that international cooperation on economic policy need not be impossible, pointing to the agreements and shared frameworks developed around nuclear weapons as evidence that countries can establish rules in areas involving major global risks.

While manufacturing could offer a short-term employment boost, Basu said India also needed to invest in its scientific capabilities, higher education and research institutions to remain relevant in an AI-driven world.

Recalling his 18 years teaching at the Delhi School of Economics, he said India's higher education system had once given the country a distinctive standing in international academic circles.

He argued that India was losing some of that influence not because it lacked talent, but because its intellectual capabilities were not being adequately nurtured or given the space to shape public policy.

An intellectually strong environment, he said, must also allow critical voices to question government decisions and identify policy failures.

“Unless you admit, unless you give voice to that, the correction doesn't happen,” he said.

Basu called for high-level technical groups to study how other countries were addressing challenges in education, research and technology. China, he said, offered lessons in the development of its higher education system and scientific capabilities.

He also highlighted the widening gap between India and China in patent activity. Citing figures he described as approximately a year-and-a-half old, Basu said China was recording around 1.044 million patents annually, compared with approximately 64,000 to 65,000 in India.

The gap, he argued, underlined the need to strengthen India's research ecosystem and examine the regulatory and institutional factors that influence innovation, including patent law and the speed of regulatory processes.

He also argued that India should develop its own large language models (LLMs), rather than assume that it could remain a bystander as the technology advances.

“If you look at them in the past, Tata Institute of Fundamental Research, Indian Statistical Institute, many of the early computer science development, visual reading by computers were being developed in India in the forefront. That's what's happening now,” he said, recalling India's earlier contributions to computer science.

For Basu, the challenge is to ensure that the country can both create new technology and prepare its workforce and institutions for the disruption it brings.

Basu challenged the familiar argument that every technological revolution ultimately creates new categories of employment to replace those it destroys.

Although earlier industrial and technological transformations had initially displaced workers before generating new forms of work, he said there was no guarantee that AI would follow the same trajectory.

“Because something has happened 10 times in the past, it will be utterly foolish to think it will happen automatically this time,” he said.

To illustrate the danger of assuming that the future will resemble the past, he invoked a story associated with philosopher Bertrand Russell about a chicken that is fed every morning by its owner, only to have its neck wrung one day. The chicken's experience, he suggested, is a reminder that a pattern repeated many times does not guarantee the same outcome indefinitely.

AI could produce a more abrupt and extensive reduction in labour demand than previous technological advances, he argued. If that happens, governments may have to consider new economic arrangements, including what he described as universal basic shares.

Unlike a one-time transfer of money, such a model would seek to give people a continuing stake in the wealth generated by productive assets and automated enterprises.

Basu also acknowledged that the prospect of a world in which machines do much of the work need not be entirely dystopian. In theory, it could allow people to spend less time working and more time pursuing music, art and other enriching activities. But such a future would be possible only if people continued to have the means to access the wealth being generated.

“It's not as dismal as it appears,” he said, describing the possibility of a society in which people could enjoy greater leisure as machines took over routine work.

The danger, he argued, would be allowing the benefits of automation to accrue to a small minority while the rest of the population lost its purchasing power.

His warning was therefore not simply about the loss of jobs, but about whether economies would be able to distribute the gains from AI in a way that allowed people to live with dignity.

For India, the immediate priority is to create jobs and strengthen its manufacturing base. But Basu's larger message was that the country must also invest in scientific talent, prepare for structural changes in employment, and contribute to the global debate on how the wealth generated by AI should be shared.