AI, Revisited | FutureIsNow
In June 2023, OpenAI CEO Sam Altman discussed one of the most consequential questions surrounding generative AI: would the technology make workers redundant, or would it make them more productive?
Speaking with Satyan Gajwani in an exclusive conversation hosted by The Economic Times, Altman argued that the early impact of AI would often be felt at the task level rather than through the immediate replacement of entire jobs. His reasoning was straightforward: AI systems could perform specific activities, but workers would still need to coordinate those activities, judge the results and determine how the pieces fitted together.
He also made a larger economic argument: if software developers became dramatically more productive, the additional output could help meet unmet demand for software rather than simply eliminate their jobs. The published ET interview report captures that argument.
More Read
It was an optimistic assessment. It was also a testable one. Three years later, the question is no longer just whether AI can help people work faster. It is whether that additional capability is translating into better output, changing employment patterns and creating enough new economic activity to offset the disruption.
What the evidence says about AI and jobs in India
India offers a particularly important setting for this question. Its IT services industry, software workforce and global capability centres are closely connected to the kinds of knowledge work that generative AI can assist or automate.
A study by the Indian Council for Research on International Economic Relations (ICRIER) surveyed 651 IT firms across 10 Indian cities between November 2025 and January 2026. Its findings do not support predictions of widespread job losses on the scale sometimes suggested in public debate. Instead, the researchers report a more complicated picture: hiring moderation, particularly at entry level, alongside productivity gains and growing demand in some technical and analytical occupations.
This matters because employment can change without a sudden wave of redundancies. A company that uses AI to improve the productivity of its existing workforce may decide to hire fewer additional people. It may also change the mix of skills it needs, increase output or expand into work that previously would not have been commercially viable.
More Read
For a new graduate entering the labour market, the result could be different from the experience of an established employee whose job is being redesigned. The ICRIER study is specific to the IT sector, so its findings cannot be generalised automatically to all Indian industries. But it offers evidence that the employment story is more nuanced than a simple choice between mass replacement and uninterrupted job growth.
Productivity gains are real. Their economic value is a separate question.
Altman’s argument depended on a second proposition: more productive workers could create more output, helping businesses meet demand that had previously gone unserved.
Indian firms report productivity benefits from generative AI. In its survey, ICRIER found that a majority of participating firms reported gains that included time and cost savings, as well as higher or better-quality output. Yet there is an important distinction between improving a task and improving a business.
A worker may draft a document more quickly, generate code faster or analyse a larger volume of information. Those improvements matter, but their economic value depends on what happens next. Does the organisation deliver more work? Does quality improve? Are costs reduced? Can the business serve more customers? Does the time saved create opportunities for employees to perform higher-value work?
The International Labour Organization’s review of empirical evidence on generative AI and productivity highlights this gap. Its international evidence finds that productivity gains are uneven and that reported time savings have not consistently translated into higher measured output, earnings or employment.
This is not evidence that AI productivity gains are imaginary. It is a warning against treating a faster task as proof of a larger economic benefit. For businesses, the meaningful measure is not simply how much time AI saves. It is what that time produces.
The entry-level question
The most consequential finding for India’s workforce may be the pressure on entry-level hiring. According to ICRIER, hiring moderation in the surveyed IT firms was concentrated particularly at entry level. That finding deserves attention because early-career roles are often where workers acquire experience, learn organisational processes and develop the judgement required for more complex responsibilities.
If AI makes certain junior tasks easier to automate or perform with smaller teams, employers may change the number and nature of entry-level positions they offer. That does not mean entry-level jobs are disappearing everywhere. Nor does it establish that AI alone caused every hiring decision. The research points to a pattern that deserves monitoring, not a universal verdict.
It also raises a longer-term question: if fewer workers enter an occupation through routine junior tasks, how will they acquire the experience needed to take on more complex responsibilities? Companies may need to redesign training, supervision and early-career work rather than assume that the next generation of experienced workers will emerge through the same pathways as before.
This is where productivity and employment policy meet. The transition is not only about whether a job exists, but also about whether workers can enter the profession, develop their skills and progress within it.
Has Altman’s prediction held up?
The evidence supports parts of Altman’s argument, but not the entire proposition. His claim that AI would often change tasks before replacing whole jobs is consistent with the more nuanced employment patterns reported by ICRIER. The study finds productivity gains and changes in hiring, rather than evidence of wholesale job destruction across the firms surveyed.
His expectation that productivity could expand economic activity is plausible, and the reported gains provide a reason to take that possibility seriously. But the evidence does not establish that increased productivity will automatically create enough additional demand and new jobs to offset every employment loss.
That distinction is especially important in India, where the scale and structure of the workforce mean that the effects of AI will vary across occupations, skill levels and sectors. The World Bank’s October 2026 India Development Update identifies significant opportunities for India to use AI to raise productivity and improve services, while emphasising that the long-term labour-market effects remain uncertain.
The most defensible conclusion is neither that Altman was simply right nor that his optimism has been disproved. His task-level explanation remains useful. The economic outcome is still being determined.
What should India watch next?
Three signals will help determine whether AI’s productivity gains translate into broader economic benefits.
- Measured output—not just time saved. Businesses should track whether AI increases the quantity or quality of work delivered, improves service, reduces costs or creates additional revenue.
- Entry-level hiring and career progression. The number of junior roles, the skills required for them and the opportunities for workers to gain experience will be important indicators of how employment is being reorganised.
- The creation of new demand. If AI makes software and other services cheaper or more accessible, businesses may be able to serve customers and pursue projects that were previously uneconomical. Whether that opportunity generates enough additional work to offset displacement remains an open question.
India’s AI story will not be decided by model capability alone. It will depend on how businesses redesign work, how workers develop new skills and whether productivity gains spread beyond a narrow group of companies and employees.
The FutureIsNow Verdict
Partially supported—with an important caveat.
Sam Altman’s 2023 argument that AI would initially change tasks and make many workers more productive is supported by evidence of productivity gains and evolving employment patterns in India’s IT sector.
But the broader promise—that higher productivity will translate into enough additional demand, better jobs and broad-based prosperity—has not yet been established.
The most important question is no longer whether AI can make a worker more productive. It is whether that productivity creates measurable economic value, who benefits from it and what happens to the people whose routes into work are changing.
Research note: The India-specific employment findings discussed here come from ICRIER’s survey of IT firms. The ILO review provides international context rather than a direct estimate of AI’s impact on India’s labour market. Neither source establishes a definitive, economy-wide causal estimate of AI’s net employment effect in India.
