AI is not taking jobs yet — but leaders must act as if it will
The New York Fed finds limited disruption. History suggests the real change comes later.
The Federal Reserve Bank of New York’s latest survey offers a striking conclusion: despite rapid adoption of artificial intelligence across industries, the technology has not yet had a material impact on jobs. Around 40% of service-sector companies and 26% of manufacturers in the district report using AI, up sharply from 25% and 16% the year before. Yet the majority of those companies are retraining staff rather than reducing headcount.
The New York Fed’s finding — that AI adoption is accelerating without large job losses — fits a familiar pattern. When new technologies enter the workplace, their effects are often gradual, reshaping tasks long before they reshape headline employment numbers.
In banking, for example, the Bureau of Labor Statistics projects that the number of U.S. bank tellers will fall by just over 8% between 2016 and 2026. The driver is clear: automation and online services have taken over routine transactions. Yet the role has not vanished. More than 170,000 people were still employed as tellers in 2024, with responsibilities that now tilt toward customer service and financial product support rather than cash handling.
Manufacturing shows the same trajectory. The OECD estimates that across its member economies, around 28% of jobs are at high risk of automation. But in practice, the shift has been incremental. Industrial robotics displaced repetitive assembly work, but also created demand for technicians, maintenance staff, and logistics specialists. The impact was not the sudden collapse of factory jobs, but the steady redesign of what those jobs involved.
It’s critical to bear in mind that the absence of mass redundancies in the early phase of adoption does not mean disruption will not come. It means the first chapter is one of augmentation, with machines handling specific tasks while the wider reconfiguration of roles unfolds over years.
AI today appears to be following that familiar pattern. Companies are applying it to narrow, repetitive processes — claims handling, invoice reconciliation, chatbots for customer queries — not wholesale restructuring. As Harry Mason, head of client services at Mason Infotech, puts it: “AI agents are not replacements for human expertise. They are supportive tools, designed to free people from repetitive, time-heavy tasks so they can concentrate on creativity, strategy, and decision-making.”
That shift is already measurable. Kevin Fitzgerald, UK managing director at Employment Hero, points to new survey data: “Employees who use AI regularly rate their own productivity 17% higher. However, productivity can decline by up to 50% in organisations where AI is implemented poorly.” The opportunity lies not in rushing to cut jobs, but in rolling out AI in a way that augments work and builds confidence.
Examples of this augmentation are emerging across industries. Gerard Donohue, CTO at Telent, notes that “AI is not replacing people, it is helping them work smarter. For most businesses, the focus is on upskilling rather than downsizing.” He points to practical cases: AI-enabled CCTV systems that assist security teams, smart buildings that cut energy waste without constant human oversight, and network monitoring that detects problems before outages occur. “These applications don’t eliminate jobs, they enhance them. The value lies in freeing up human capacity, reducing fatigue, and improving response times,” Donohue says.
