It sounds dystopian, but it is not unreasonable to suspect that the rise of artificial intelligence is not just another global warfront, but the beginning of a race to make human labour largely redundant over time. AI ultimately threatens to eliminate most kinds of work that we currently aspire for.
The Economic Survey 2024-25, presented to parliament on 31 January, has one chapter—chapter 13, titled Labour in the AI Era: Crisis or Catalyst?—that raises deep concerns about AI. This is despite striking a hopeful note that India could use the window of opportunity between now, when AI isn’t quite a job-destroying Godzilla, and when it will become so, to create social institutions and skilling opportunities to benefit from this technology.
Says the Survey: “The increasing complexity of the models being developed today represent a paradigm shift in the field of AI, showing the world that in a few years, ‘intelligent machines’ will be capable of performing tasks that are predominantly handled by humans today….”
The survey is right to flag this concern, though it is unclear whether India will get its act together before AI starts chewing up good quality jobs.
The countdown to AI armageddon began with OpenAI launching ChatGPT in November 2022. It got over 100 million downloads within two months. Google then launched Gemini, which is now a permanent fixture in all Google searches and Android phones. Last week a Chinese firm based out of Hangzhou launched DeepSeek for free download on Apple’s platforms, and it shocked the market with its versatility and frugal use of capital and hardware. It quickly became the top download on Apple’s iOS. US tech stocks crashed when the market saw the gap in performance between OpenAI and DeepSeek. Last Monday saw a Nasdaq crash, led by a more than half-a-trillion-dollar drop in chipmaker Nvidia’s share valuations. Normally, a chipmaker should see demand rising if a new tech platform encourages more usage, but DeepSeek’s frugal use of chip capacity dented Nvidia, though that will change once more entrepreneurs board the AI bandwagon. A few days after DeepSeek, another Chinese company, Alibaba, announced its own new AI model, Qwen 2.5Max, which claims to be better than DeepSeek. Barely two weeks ago, the US announced the Stargate project involving a $500 billion investment by tech giants to build a massive AI infrastructure by 2029.
India formally joined the AI race a few days ago when IT Minister Ashwini Vaishnaw announced that over 18,000 graphics processing units (GPUs) will be offered at a subsidised rate by 10 empanelled companies to provide the processing power for developing AI models and applications. He expects India-specific foundational AI models within the next 10 months. GPUs, which allow for multiple tasks to be processed in parallel, were originally designed for computer graphics, but AI is where they are proving most useful and versatile.
With big bucks now in play, the pace of AI launches and adoption could become truly breathtaking in the coming months. Even though current AI models are hardly doing enough to directly threaten jobs too much, the problem should be obvious: if the rise of the mobile internet and automation up-ended middle-level jobs in spheres like telecom, banking and financial services, AI will threaten the upper-end of the job market in addition to the middle levels. If AI can code, count, calculate, reason and evaluate, not just lower-skilled software programmers, but accountants, data analysts, legal and media professionals also need to worry, says one CBS report. Google’s Sundar Pichai says AI is going to be a bigger game-changer than the internet itself.
Also read: India solves one problem, adds another while using AI for climate crisis. Clean up data first
Who bears the brunt?
This disruption will not happen immediately, but in stages, for workplaces have to adapt to the new AI tools available to them and train their people to use them. But once that process is underway, only jobs that require super skills will be safe. For example, it was once thought that computers could not do jobs that even middle-skilled journalists or designers could do. Now, generative AI can do most of the work of journalists and graphic designers. Editing, rewriting, paraphrasing, summarising, picture and image generation and management are just a minute’s work.
The medical profession will be impacted as most diagnoses and evaluations can be done using patient databases and technology. India’s doctors will soon find that only super-specialisations will enable them to earn enough to pay for their high-cost education. Medical consultations will become a commodity requiring no specialised knowledge. Logically, what general physicians do today should be doable, with the aid of technology, by nurse practitioners and physician assistants with shorter-term courses, says Dr Suneel Dhand, a US-based physician. This will force a complete rethink of regular medical courses since core medical knowledge can now be embedded in chips and databases.
Elite education, where people pay an-arm-and-a-leg for tuition by incurring deep debts, may also be the next big business to tank. Why spend $80,000 or more annually for a four-year degree in the US when even top Ivy League B-schools (with two-year courses) are unable to get jobs for 16 per cent of their alumni? An Economist report says that in 2024, the share of students landing a job three months after graduating fell by six per cent to 84 per cent. And if AI is going to dent the market values of more such elite qualifications, the future of high-paying jobs looks bleak.
AI optimists will tell you that disruptive technology and innovations usually create more jobs than they destroy as new uses for technology arise. That may well be true, but two questions arise: how fast will AI destroy jobs and how fast can the losers be upskilled? The time gap matters. Secondly, if an ever-improving AI capability is sooner or later going to match or exceed human mental and physical capabilities, how can skilling alone help? What if AI masters human skills faster than humans?
India, with its millions of job-seekers, is at the wrong end of the AI expansion curve. In countries with declining populations (Europe, China, Japan), AI can actually help by substituting for a shrinking workforce. That is not the position India is in, as its labour force is growing and foundational skills—the ability to read, write and count—are lacking in large sections of the population.
Also read: India’s tech bros’ new fad is Indic AI — Krutrim to AppsForBharat. The answer to ChatGPT
A race you can’t lose
Unfortunately, it is a race nobody can afford to lag behind in or lose.
Just hark back to 1945, when the US exploded its first atomic bomb. Once the fear of nuclear annihilation gripped everybody, the Soviet Union followed suit with its own bomb in 1949. Britain (1952), France (1960), China (1964) and India (1974) came along later. In 1957, when the Soviet Union launched its first satellite (Sputnik-1), the shock forced the US to spend billions of dollars on space tech and ballistic missiles.
The DeepSeek launch is triggering a massive flow of investments into AI, no country wants to be left behind in the race for super intelligence capabilities. The tragedy is that this race is ultimately about making human labour less and less relevant to the production of goods and services in the modern economy. That cannot be good news.
There is little doubt that AI will be enormously beneficial to those with access and the ability to use it. Pretty soon AI agents can be used by us to do practically every one of our day-to-day tasks, from replying to email to ordering groceries to booking tickets to the Kumbh or Coldplay concerts. It will also make education and health accessible to a wider range of people at low costs. But AI will also make it easier for scamsters to scrape your personal information from the web and break your passwords and codes easily.
The biggest negative impact may be on worthwhile jobs. The Economic Survey quotes global studies to point out potential job losses. It says: “The International Labour Organisation estimates that nearly 75 million jobs globally are at complete risk of automation due to AI….Estimates from private sector firms paint a similar picture. Goldman Sachs economists state that nearly 300 million full-time jobs remain exposed to AI-driven automation. McKinsey estimates demonstrate how, by 2030, up to 30 per cent of current work-hours could be automated by generative AI15 across Europe and the United States”.
The survey makes for depressing reading.
The nuclear arms race was about the potential to destroy human lives. We have survived that by adopting the strategy of mutual assured destruction (MAD), which means all nuclear powers have an incentive to avoid using this mass killing technology.
The AI race is about destroying livelihoods. And there is no MAD strategy available to prevent the rapid destruction of jobs.
R Jagannathan is editorial director at Swarajya magazine. Views are personal.
(Edited by Theres Sudeep)