AI agents are now performing the tasks that used to launch careers, compressing the job ladder and raising the risk of a white-collar jobless recovery.
The entry-level job is disappearing quietly. Agentic AI systems now absorb the multi-step tasks across coding, writing, research, and customer support that once defined the first two years of a knowledge worker's career. This is not factory-floor automation. It is a structural compression of the job ladder, starting at the bottom rung.
Goldman Sachs Research estimates 300 million jobs globally are exposed to AI automation, with AI capable of handling 25% of all US work hours. Economist Joseph Briggs projects 6 to 7% of workers will face displacement, with entry-level knowledge and content workers identified as most immediately at risk.
Traditional automation replaced repetitive physical tasks. Agentic AI operates differently, executing multi-step goals autonomously across writing, research, coding, and customer support without human direction at each step.
Marco Argenti, Chief Information Officer, Goldman Sachs
Goldman Sachs CIO Marco Argenti told Fox Business the shift is already complete:
"We used to look at models as a chat that would provide questions and answers. Now a year later, we look at models as essentially entities or agents that can perform tasks on your behalf."
Argenti predicts 2026 will see the rise of "agent as a service," where companies rent fleets of AI agents specializing in coding, finance, customer service, and design instead of onboarding junior employees. For a hiring manager weighing the cost of a junior analyst against an AI agent that never needs onboarding, the math is becoming difficult to ignore.
The structural problem is not just job loss. It is the elimination of the training ground. Entry-level roles have historically served a dual function: producing output for the firm and producing capability in the worker. Agentic AI is absorbing that output function and removing the on-ramp entirely.
Goldman Sachs projects AI displacement concentrated in knowledge sectors.
Goldman Sachs Research finds that workers displaced from knowledge industries are particularly mismatched for the roles the economy is actually creating, including construction, electrical work, HVAC contracting, and data center buildout. Construction jobs tied to data center expansion have increased by 216,000 since 2022, and roughly 500,000 net new infrastructure jobs will need to be filled by 2030. These are not roles a displaced junior content strategist or entry-level analyst can step into without significant retraining.
The jobless recovery risk in white-collar sectors is no longer theoretical. Goldman Sachs projects unemployment rising to 4.5% in 2026, with AI job losses potentially pulled forward faster than the 10-year base case assumes. If that acceleration happens, the Federal Reserve may be forced to cut rates to offset the drag.
Adaptability is now the most valuable workplace skill, but adaptability requires opportunity. Workers who never get hired into entry-level roles cannot adapt their way into mid-career expertise. The policy and workforce development response to agentic AI needs to move as fast as the technology itself, or the next generation of knowledge workers will find the ladder removed before they had a chance to climb it.