There is a familiar argument about artificial intelligence that goes something like this:
“Relax. Technology has always destroyed jobs and created new ones. The horse-and-buggy industry disappeared, but automobiles created millions of jobs. Computers eliminated some clerical work, but they also created entire industries.”
That argument is not wrong. It is simply incomplete.
The immediate economic problem with AI may not be that it eliminates the job of a 45-year-old manager. It may be that it eliminates the first job a 22-year-old needs in order to become a 45-year-old manager.
That first rung on the career ladder is already being sanded down.
The entry-level job market is where the pressure is showing up
Goldman Sachs recently analyzed employment growth across more than 800 occupations. The broad conclusion was not that every job is about to vanish. The more precise conclusion was more uncomfortable: AI-related employment headwinds are strongest among entry-level workers.
Across the broader labor market, a 10% increase in occupational exposure to AI was associated with roughly a 0.1 percentage point drag on annual employment growth in the United States, Canada, and France.
For entry-level workers, the drag was more than twice as large in the United States and exceeded 0.6 percentage points in Australia.
That may sound like a small statistical difference. It is not small when multiplied across millions of young people trying to get their first foothold.
A career is not built by teleportation. Most people begin with work that is repetitive, supervised, and not especially glamorous:
- Answering customer-service calls
- Updating spreadsheets
- Reviewing documents
- Scheduling appointments
- Entering data
- Preparing basic reports
- Writing routine marketing copy
- Supporting a more experienced employee
- Handling the first draft before somebody else handles the final one
These jobs have always been the apprenticeship system of the modern office. They teach judgment, communication, reliability, and the unwritten rules of professional life.
Artificial intelligence is particularly good at doing the repetitive part.
The call center is the warning light
Call centers offer one of the clearest examples of what is happening.
Goldman’s research found that call-center employment is now approximately:
- 39% below its long-run trend in the United States
- 33% below trend in Canada
- 27% below trend in Germany
That does not mean AI caused every missing job. Businesses over-hire, under-hire, restructure, outsource, and occasionally make decisions that have nothing to do with technology. Economic data is not a courtroom confession.
But the pattern is hard to ignore. Call centers are precisely the kind of business where an automated system can answer common questions, classify customer problems, summarize interactions, and route complicated cases to a smaller number of human workers.
A company that used to need 100 people to handle basic customer contact may now need 60 people supervising automated tools and handling exceptions.
The customer may still reach a human eventually. The young worker may never get hired in the first place.
That is the distinction being missed in most of the loud arguments about AI and employment. Automation does not need to replace every worker to change the labor market. It only needs to reduce the number of openings available to beginners.
Tech layoffs are not the entire story: but they are a loud one
Technology companies announced 149,023 layoffs through July, according to Challenger, Gray & Christmas. That was up 67% from the same period last year, and technology accounted for approximately 31% of all announced job cuts in 2026.
Some of those layoffs are clearly connected to AI. Others reflect the hangover from pandemic-era over-hiring, high interest rates, mergers, shifting business models, and executives discovering that the phrase “do more with less” sounds better in a boardroom than it feels in a household.
Still, AI is becoming a convenient and increasingly capable excuse for shrinking headcount.
The young worker is vulnerable because employers have a choice. They can hire three junior employees and train them over time, or they can purchase software that handles much of the routine work immediately. In a difficult economy, the software does not ask for health insurance, vacation time, a raise, or a desk.
That does not make the software evil. It makes the incentives obvious.
The honest counterargument: new jobs are being created
The argument should not be reduced to “AI bad, humans good.” That is a slogan, not an analysis.
Hiring plans through July reached 107,500, up 25% year over year, with some of the strongest demand appearing in aerospace, energy, and manufacturing. Those industries need technicians, engineers, electricians, construction workers, maintenance specialists, and people who can operate in the physical world.
Goldman has also pointed to the growing demand created by data centers and the infrastructure needed to power them. Its research estimates that roughly 500,000 additional workers could be needed in the United States by 2030 to meet rising power demand tied partly to data-center construction and related infrastructure.
That is real opportunity. It is also a reminder that the labor market does not automatically move people from a displaced office job into a high-paying industrial or technical job.
The person who lost a junior accounting position may not be qualified to install electrical systems. The call-center worker may not have the training to maintain a cooling system at a data center. A recent communications graduate cannot simply walk into an aerospace plant and start certifying aircraft components.
The economy may create jobs and still leave millions of people stranded between the old ones and the new ones.
AI is most useful when it augments a worker
The most encouraging evidence comes from workplaces where AI assists people instead of replacing them.
A nurse using software to summarize patient notes is still a nurse. An engineer using AI to test design options is still an engineer. A manufacturing technician using computer vision to identify defects may become more productive and more valuable.
The Goldman Sachs analysis of the U.S. labor market makes this point clearly: the effects are concentrated in specific industries and occupations, not evenly spread across the whole economy.
Where AI handles the paperwork and lets a human spend more time on judgment, relationships, and difficult problems, employment can hold steady or rise.
Where AI handles the entire first layer of work, entry-level opportunities are more likely to disappear.
That difference matters because junior workers need practice. A company cannot have an office full of senior employees forever if it stops hiring juniors today. At some point, the pipeline runs dry. The experienced manager retires, the senior engineer leaves, and nobody is prepared to take over because the apprenticeship jobs were eliminated a decade earlier.
Corporate America may be automating the very training system it will later complain does not exist.
The first job is more than a paycheck
A first job teaches young people things that do not appear on a résumé:
- How to show up when the weather is bad
- How to deal with an unreasonable customer
- How to ask a useful question
- How to accept criticism without collapsing
- How to finish an assignment that is boring
- How to work with people who think differently
- How to understand what a business actually does
These lessons are difficult to automate because they are learned through contact with other human beings.
When the first job disappears, the damage is not limited to a smaller paycheck. It can delay independence, reduce confidence, interrupt skill development, and make the next job harder to obtain. Employers often demand experience for positions that used to provide the experience.
That is how a labor market develops a locked door at the bottom.
What parents, workers, and employers should watch
For young workers, the answer is not to avoid technology. It is to avoid being defined by tasks that software can perform with a click.
The durable advantages are judgment, technical competence, communication, physical-world skills, relationship management, and the ability to use AI without blindly trusting it. The worker who can supervise a system, catch its mistakes, explain its output, and apply it to a real business problem is in a much stronger position than the worker who only performs the routine task the system was built to replace.
For employers, the short-term savings from eliminating entry-level roles may create a long-term talent shortage. Training costs are not waste. Sometimes they are the price of having competent employees in five years.
For policymakers and schools, the uncomfortable question is whether education is preparing young people for work that exists: or merely issuing credentials for jobs that are being redesigned.
The transition will not be painless. Goldman estimates that 6% to 7% of workers could be displaced during a broad AI adoption cycle, although the long-term impact may be moderated by new jobs and higher productivity. Historically, technology has created enormous numbers of jobs. It has also created enormous numbers of people who had to start over.
Both facts can be true at the same time.
AI may not take your job this year. It may not take your child’s job either. But the first-job squeeze is already visible in call centers, junior technology roles, administrative support, and routine white-collar work.
The regular-guy question is not whether artificial intelligence will make the economy more productive. It almost certainly will.
The question is whether the productivity gains will create a better ladder: or simply remove the bottom rungs and tell young people to climb anyway.
Disclosure: Regular Guy Economics is not a financial advisor. This article is for educational and informational purposes only and is not investment advice, a recommendation to buy or sell any security, or a guarantee of future economic or market outcomes.
Be mindful, be watchful and good luck.