Georgieva Warns AI Is Pulling Away Graduates’ First Rung

IMF Managing Director Kristalina Georgieva said on October 12 that around 40% of jobs worldwide face AI. She called the change a “slow-moving tsunami” already hitting young workers.

The same boom, she said, is holding world growth near 3% even as an energy shock drags. Hiring is moving into lower-paid construction, data-center and service work, and away from the routine posts that used to be a graduate’s first rung.

A Slow Tsunami Opened the Bangkok Meetings

Georgieva was speaking at the opening flagship panel in Bangkok, on the first morning of the IMF and World Bank annual meetings at the Queen Sirikit National Convention Center. The session, “Small AI: More and Better Jobs,” put her beside World Bank Group President Ajay Banga and Singapore President Tharman Shanmugaratnam, who co-chairs the Bank’s Jobs Council. The meetings run through October 18. The hall first hosted these gatherings in 1991.

She told the room the technology is already a demand shock and an investment shock. In her account, that extra gear is what is keeping the world economy growing at around 3%, and it could add 0.5 to 1 percentage point to growth in advanced Asian economies and 0.2 to 0.8 points in emerging Asia. Hardware builders, including in Thailand, are taking the first lift from data-center construction.

Well, it is a slow-moving tsunami, but a tsunami nonetheless. The impact over time on jobs globally, we estimate to be around 40% jobs either augmented or downgraded or eliminated.

Kristalina Georgieva, Managing Director, International Monetary Fund, Bangkok panel, October 12, 2026

A slow wave is easy to live with until the water is at the door. That is the political hazard she was describing: disruption that looks gradual enough to postpone, then rearranges how people enter work and how states collect tax.

The 40% Figure Includes Jobs That Get Better

Georgieva’s live split was blunt. Around 40% of jobs worldwide, about 60% in advanced economies, and 23% in low-income countries. She waved off the idea that poorer countries are safer. “You may say, ‘Oh, they’re better off.’ No, that is just because AI is further away,” she said. Small AI, in her telling, has not reached all of them yet.

Those shares come from Fund research first set out in a January 2024 staff discussion note. That paper put almost 40 percent of global employment in the exposed group, with about 60 percent of jobs in advanced economies and 40 percent in emerging markets. It also said women and college-educated workers are more exposed and better placed to gain, while older workers may find it harder to adapt. About half of the exposed jobs in rich economies, the authors judged, could be complemented by the tools rather than replaced by them.

AI JOB EXPOSURE, AS STATED ON OCTOBER 12

Exposure is not a layoff notice. A job counted in the 60% can pay more if the worker uses the tools, or it can shrink into a worse version of itself. Young college-educated staff sit in both columns at once: the 2024 note called them the most vulnerable and the most able to switch.

The First Rung Is Already Loosening

The people to worry about “first and foremost,” Georgieva said, are young people coming into the labor market. They used to take routine, repetitive jobs and then grow up the ranks. “But if they’re gone, I don’t think any young person wants to be a barista in a coffee shop in the professional future.”

That line is doing more work than a joke about lattes. The Fund’s own U.S. slice, which she walked through on the panel, found that about 10% of jobs already require higher AI skills and pay better. Routine posts are shrinking. Low-paying work is rising in two directions, construction for data centers and services. Net employment in the area studied was positive. “That’s not a story to celebrate,” she said, “because it is positive on the side of lower-paying jobs.”

She then told the room to burst the myth that AI is shrinking demand for people. “What it is doing is it is shifting demand for people now. That may not be the case in the future.”

Shanmugaratnam, speaking after her, called the first job after graduation the place the damage shows up first. Demand for that rung is already weaker, he said, and the pattern is clearest in the most AI-exposed sectors, starting with tech. If a class waits a year and then collides with the next class, the loss is more than a delay. “You lose more than one year of skills,” he said. “You lose a whole set of intangible attributes, and you lose trust in the future.”

U.S. flows match that description more closely than a mass-unemployment chart would. The Burning Glass Institute, in an October 6 note, found that an unemployed 22- to 26-year-old with a bachelor’s degree had only a 23% chance of being employed the next month, averaged over two years, the slowest graduate job-finding pace since 2003. After the Great Recession the same measure bottomed at 24.5% in early 2011. Layoffs are not the main machine. Hiring into the old starter roles is.

WHERE EXPERTS DISAGREE

• Kristalina Georgieva: AI is already shifting demand and will, over time, touch around 40% of jobs, with young workers first in line.
• Sam Altman: The OpenAI chief said in May he was delighted to have been wrong about how fast entry-level white-collar jobs would vanish.
• Dario Amodei: The Anthropic chief has described a world in which 90% of a job is automated and the remaining 10% expands into the whole of what people do, at much higher output.
• Donald Trump: The U.S. president said in January that AI, which he wants called “super intelligence,” will be a tremendous job producer, with robots filling roles people cannot.

Altman’s admission and Georgieva’s warning can sit in the same year because the wave is slow. Jobs remain. The training posts that used to turn a graduate into a mid-career worker are the ones getting thinner.

Singapore’s 2025 Graduates Found Fewer Full-Time Jobs

Shanmugaratnam did not need a forecast to make the first-rung point. Singapore’s six autonomous universities have already printed the slide. In the 2025 Joint Autonomous Universities Graduate Employment Survey, 74.4% of labour-force graduates were in full-time permanent jobs about six months after final exams, down from 79.4% in 2024, a drop of 5.0 percentage points. Some 88.9% had secured some form of employment. About 14,400 graduates responded, 73.7% of the cohort, with status measured as at November 1, 2025.

Pay for those who landed full-time work did not crack. Median gross monthly salary held at S$4,500, the same as in 2024. More people tried to enter the market anyway. The labour-force share rose to 92.2% from 90.7%. The share who applied and got no offers rose to 8.5% from 5.7% in 2024 and 4.1% in 2023. Part-time or temporary work rose to 7.2% from 6%.

SINGAPORE GRADUATE SURVEY, 2024 TO 2025

Singapore’s education ministry told parliament in April that 2022 and 2023 were a post-pandemic hiring surge that was always going to cool, and that 2025 sits closer to the pre-COVID pattern. Vacancies have moderated, the ministry said, amid cautious hiring and “significant economic and geopolitical uncertainties.” AI “may, in part” be reshaping how work is done. “The specific impact of AI on the demand for entry-level jobs remains uncertain at this point.”

The steepest full-time drop was in Arts, Design and Media, down 12 percentage points from 2024. Count freelance, project and contract work and the decline shrinks to 4.1 points. Voluntary part-time and temporary work in that cluster rose 5.5 points. In October 2025 the government launched Graduate Industry Traineeships, GRIT and GRIT@Gov, as a bridge into industry experience, then extended the schemes to the 2026 cohort.

On a 10-year view, the ministry still puts the picture at around 9 in 10 graduates securing employment within six months. The argument on the Bangkok stage was about the quality of that first foothold, not a collapse in the headline employment rate.

Construction and Services Took the New Hiring

Put Georgieva’s U.S. slice next to the Singapore survey and the shape repeats. People are still being hired. They are being hired into different work, often at the bottom of the pay scale, while the old junior-office intake stalls.

WHAT THE U.S. SLICE ALREADY SHOWS

• AI skills premium: About 10% of jobs in the Fund’s U.S. study already require higher AI skills and pay better.
• Routine work: Repetitive posts are shrinking, which is the layer graduates used as a first step.
• Where headcount rose: Lower-paying jobs increased in data-center construction and in services.
• Net jobs: The studied area still added employment, concentrated on the low-pay side.

Cheering the death of “pointless” routine work, a line that ran through replies to the IMF remarks, skips the training function those posts had. If the junior analyst role is the one a firm can most easily leave unfilled, the senior role does not train its own replacement. Shanmugaratnam called the risk a K-shaped labor force, with the broad middle falling out, unlike the first automation wave that hit the low end of manufacturing.

He also said the silver lining is speed. Displacement, he argued, arrives task by task, which leaves time to learn on the job while people still hold posts. “AI, as a tsunami, isn’t going to come all of a sudden,” he said. “It’s going to come gradually.” That is the same slow clock Georgieva used, and it is why a 3% growth print can hide a broken intake.

What Do Governments Tax When Labor Shrinks?

Georgieva’s third ask was the one that sounded most like an IMF managing director. Study the labor market, she said. Rethink education so a skill is not a lifetime possession. Then look at the budget.

“Traditionally, governments would generate revenues by taxing, mostly, people. Well, if AI changes the structure of the labor market, and then there is less labor, more capital, what do we tax?” If revenues shrink, she said, governments will be in a tough spot just when they are supposed to help society adapt.

THREE MOVES SHE ASKED FOR IN BANGKOK

• Measure the shift: Gather and analyze labor-market data instead of arguing from slogans, which is the work she said the Fund is doing.
• Rebuild education: Treat lifelong learning as a necessity, because a single skill stack no longer lasts a career.
• Reopen the tax question: If income moves from wages toward capital, finance ministries need another way to fund the adjustment.

That fiscal bind is the slow wave’s second hit. A state that lives off wage taxes will find it harder to pay for the traineeships Singapore is already running, or for the crisis-era hiring supports Shanmugaratnam wants kept on the structural menu, the tools used after 2008 and during COVID. He said those supports should no longer be treated as emergency kit.

Georgieva closed her run with a limit. “We cannot today imagine what exactly the world of work will be five years from now.” The instruction she attached was to measure what is already moving and act on it.

Developing Economies Are Racing a Different Clock

Banga spent his time on a larger intake problem. In the next 15 years, he said, 1.2 billion young people will turn 18. That can be a dividend if they grow up with clean air, water, health care, schools and skills, then find a way to be productive. Governments enable, he said. The private sector generates between 75% and 85% of jobs in any country.

In emerging markets, he argued, the live question is not giant language models. It is applications that raise output in a school, a clinic or a farm. He described an illiterate woman farmer in India pointing a 3G phone at the back of a leaf, matching a disease to an insecticide, and paying 25 rupees at her cooperative because the country put bank accounts in everyone’s hands. “When we hear that AI is coming for our jobs, that is not necessarily the case in the developing world,” the session’s own film said. “There, the fundamental risk is not that AI takes jobs; it is that we are too slow to use AI to create them.”

Shanmugaratnam said growth in China, India and developing Southeast Asia is already running ahead of job creation, and that passivity on AI will widen that gap. He still sees room in manufacturing where labor is cheaper than China and cheaper than robots, in green power that he said yields three times as many jobs as fossil fuel energy for the same billion dollars, and in health systems where nurses per capita run from one-fifth to one-tenth of rich-country levels. He also pointed to one-person firms using AI to reach markets, a form of work with more agency than a missing junior desk job.

The Bangkok meetings still have days to run. The warning from the opening morning is already specific enough to use. AI is paying for this year’s growth. It is also taking apart the first step that used to turn a new graduate into a taxpayer with a career, and the invoice for that shift lands on the same governments being asked to fix it.