Artificial intelligence is no longer just changing how people work. It is now beginning to reshape which jobs grow, which shrink, and which workers are left struggling to find a foothold in the economy.
New research by Goldman Sachs shows that AI is creating a sharp divide across the labour market. Some professions are seeing stronger hiring because AI makes workers more productive. Others are already facing pressure as companies use AI tools to automate repetitive tasks once done by humans. The findings suggest that the impact of AI is no longer theoretical. It is already visible in hiring trends, payroll growth, and unemployment numbers across the United States.
Goldman Sachs economists estimate that AI reduced monthly payroll growth by around 16,000 jobs over the past year and slightly increased unemployment. At the same time, jobs where AI acts more like an assistant than a replacement added roughly 9,000 payroll positions a month. The split reveals an important reality about the AI economy: not all jobs exposed to AI are equally at risk.
The clerical and repetitive jobs under pressure
The research shows the biggest substitution threat falls on occupations built around routine and predictable tasks. Telephone operators rank as the profession most exposed to direct AI replacement. Insurance claims clerks, bill collectors, billing clerks, telemarketers, payroll clerks, legal assistants, procurement clerks, proofreaders, and word processors also appear among the most vulnerable occupations.
A clear pattern runs through the list. Nearly all these jobs involve repetitive communication, documentation, data entry, form processing, or scripted interactions—exactly the type of work generative AI systems now perform quickly and cheaply. The graphs show telephone operators sitting far above other occupations, this also tells us how voice automation and AI-powered customer handling systems are advancing.
Even clerical back-office roles that once seemed stable are now appearing vulnerable. Billing clerks and payroll clerks, for example, depend heavily on structured workflows that AI can increasingly handle with limited human oversight.
Proofreaders and word processors are also among the most exposed jobs, indicating how language-based AI systems are disrupting traditional administrative and editorial support work. The study notes that occupations facing stronger AI substitution are already seeing declines in job postings and reduced operating costs as firms automate tasks.
AI is strengthening high-skill professions
But the research also shows another side of the AI story, one where technology boosts workers instead of replacing them. Education administrators top Goldman Sachs’ list of occupations most likely to benefit from AI augmentation. Construction managers, chief executives, physicians and surgeons, first-line supervisors, lawyers, airfield operation specialists, industrial production managers, operation managers, and industrial engineers also rank highly.
Unlike clerical jobs, these professions involve leadership, strategic decisions, human coordination, creativity, and physical-world problem solving areas where AI still struggles to operate independently. For lawyers, AI may help analyse contracts and speed up research, but courtroom strategy, negotiation, and judgment still require human expertise.
For doctors and surgeons, AI can assist in diagnosis and data analysis, but patient interaction, trust, and critical medical decisions remain deeply human responsibilities. Construction managers also stand out because their work combines planning with physical site coordination — something AI cannot fully automate.
The jobs safest from automation may not be the most technical
One of the report’s most striking insights is that jobs do not become safer simply because they are highly skilled or technology-heavy. Instead, occupations that combine human judgment, interpersonal interaction, creativity, leadership, and unpredictable real-world conditions appear more resilient.
The report gives the example of customer service workers and interior designers. Both roles are exposed to AI tools, but interior design requires creativity, unstructured thinking, and physical interaction with spaces and clients. That makes the profession harder to replace completely. This distinction is becoming increasingly important as companies decide where AI can cut labour costs and where humans still add irreplaceable value.
Young workers are being hit first
The research suggests younger workers may be carrying the heaviest burden from the early AI transition. Many entry-level jobs traditionally relied on administrative and repetitive tasks, the same tasks AI now performs efficiently. As companies automate those responsibilities, fewer openings may remain for fresh graduates and inexperienced workers trying to enter industries. That could partly explain why competition for basic entry-level roles has intensified even as companies continue investing heavily in AI systems.
The report suggests that future hiring may increasingly favour workers who can manage AI tools while also offering communication skills, adaptability, leadership, and creative thinking.
Why AI may still create new demand
Despite concerns around automation, Goldman Sachs says the long-term picture may not be entirely negative. The economists point to a concept called the “Jevons paradox,” where greater efficiency can sometimes increase overall demand instead of reducing it. As AI lowers costs and boosts productivity, businesses may expand services, produce more output, and create new forms of demand that eventually generate additional jobs. That pattern is already visible in occupations with strong augmentation potential, where companies are seeing productivity gains alongside rising hiring activity.