At the turn of the millennium, India’s medical transcription industry looked like proof that globalization could build durable middle-class careers. Fiber optic cables, English fluency, and a favorable time-zone gap turned Bengaluru into an overnight processing hub for American hospital dictations. Young workers trained for months, companies built academies, and recruiters made promises about long-term futures. Within a generation, that entire architecture is being dismantled, not by a recession or a policy shift, but by automated platforms that do the same work faster and cheaper. This is not a story about one industry. It is a story about what happens when the cognitive core of professional life becomes automatable.
The Jobs That Vanished Overnight
The speed of disruption is what makes this wave different from previous ones. In Bengaluru, American clients began canceling contracts in early 2024, shifting work to automated platforms. Hundreds of jobs disappeared in a matter of weeks. The cafeteria that once operated at 3 a.m. went quiet. Hiring froze. The recruitment team itself was let go. New hires were brought in under “conditional retention training” arrangements, only to be dismissed shortly after, a practice the source describes as an accounting maneuver to inflate headcount figures for prospective clients.
The same pattern is visible across geographies. In Manila, tens of thousands of transcriptionists have been displaced. In Nairobi, call center operators now compete directly with chatbots. In Colombia, customer support roles are being absorbed by generative systems. Economists describe these positions as “high exposure, low complementarity”: roles where automation replaces the worker rather than amplifying what the worker can do.
What makes this structurally significant is the type of work being eliminated. Earlier waves of mechanization targeted manual labor. This one targets knowledge work, the codified, language-based, judgment-adjacent tasks that powered the outsourcing boom and built middle-class livelihoods across the Global South.
The Numbers Behind the Disruption
The scale is not abstract. Between 2022 and April 2024, India’s tech sector shed more than 500,000 jobs, with 425,000 layoffs occurring in 2023 alone. Net hiring across India’s top five IT companies in the first nine months of 2025 came to just 17 positions. The former CEO of HCL Technologies has warned that as much as 70% of IT roles could disappear. Globally, independent trackers reported roughly 122,500 tech layoffs across 257 companies in 2025, with broader analyses putting the worldwide figure above 244,000, frequently citing AI-driven efficiency gains as a contributing factor.
The exposure is not limited to entry-level or low-skill roles. A 2026 study by Anthropic researchers, measuring actual AI usage patterns against Bureau of Labor Statistics employment projections, found that occupations with higher observed AI exposure are projected to grow less through 2034. The workers most at risk are disproportionately female, more educated, and higher-paid: precisely the people who were told their qualifications would insulate them. In five of the six countries studied, women face higher displacement risk than men.
The risk extends into fields that once seemed structurally protected. The source cites exposure figures of 84% for bioengineers, 80% for mathematicians, and 72% for editors. Creativity and complexity, long considered the last line of defense against automation, are no longer reliable shields.
What “Growth Without Work” Actually Means
Here is what most coverage of AI productivity misses. GDP growth and job creation moved in rough tandem for most of the 20th century. More output meant more employment. That relationship has broken down. Economists at the OECD note that generative AI could add up to 6.4% to GDP in advanced economies while simultaneously displacing millions of workers. The phrase they use is “growth without work,” and it describes something more troubling than a temporary adjustment: an economy where output expands, corporate margins widen, and the gains do not translate into either employment or wages for the majority.
An IMF paper cited in the source calls this the “Great Divergence.” Capital flows toward the economies best positioned to deploy AI and robotics. Developing countries face temporary GDP declines and long-term terms-of-trade losses. In high-income economies, roughly 60% of jobs show significant generative AI exposure. In low-income economies, the figure is closer to 26%, which sounds like protection but actually means fewer opportunities to leverage new tools while still absorbing the downstream effects of falling wages and displaced industries.
The historical parallel the source draws is instructive. In September 1945, more than 15,000 elevator operators in New York City walked off the job, briefly paralyzing the city’s towers and government offices. Within a few years, manufacturers redesigned the technology. By 1950, Otis had installed the first fully automated elevators. By the 1970s, the operators had vanished entirely. The difference between that story and this one is scope. Elevator operators were one profession. What AI is hollowing out is an entire class of work, across sectors, geographies, and education levels, simultaneously.
In Short
AI is not just automating tasks. It is restructuring which kinds of work generate stable livelihoods and where those livelihoods exist. The productivity gains are real. The GDP growth is real. What is also real is that the costs fall hardest on the young, the clerical, the feminized, and the Global South, while the benefits concentrate elsewhere. If current trajectories hold, the source projects automation equivalent to 300 million full-time jobs. The elevator operators took decades to disappear. This time, the cafeteria empties in weeks.
Based on reporting from Rest of World.