Customer service representative roles are contracting at an accelerating pace, with employment falling 4.8% year-on-year through May 2025—a sharp divergence from the broader labour market's 0.8% growth. This decline sits within a wider pattern affecting 18 AI-exposed occupations accounting for roughly 10 million jobs, which collectively shed 0.2% of positions whilst the rest of the economy expanded. Excluding medical roles (buoyed by healthcare sector growth), the remaining 17 occupations contracted 1.6% for the second consecutive year. The data suggests AI deployment is no longer theoretical; it's reshaping employment architecture in real time. Goldman Sachs research distinguishes between roles facing substitution—where AI replaces human labour outright—and those suited to augmentation, where AI enhances rather than displaces workers. Customer service sits squarely in the substitution category, making the 130,000-job loss particularly significant for CX teams already operating within platforms like Zendesk and Freshdesk that now embed agentic capabilities.
The question facing support leaders is not whether AI will displace headcount, but how quickly and at what scale. Since ChatGPT's November 2022 launch, credit authorizers have contracted 26.2%, broadcast roles 20.8%, and sales engineers 13.2%—trajectories that suggest customer service's current 4.8% decline may accelerate as AI agents mature. For teams currently evaluating or deploying Salesforce Agentforce or similar autonomous systems, this data validates the business case for automation whilst simultaneously raising staffing strategy questions: are you rightsizing teams toward higher-value work (coaching, complex problem-solving, relationship management), or simply reducing headcount? The distinction matters operationally and ethically. Organisations treating AI as pure cost-cutting will face talent retention issues and skill atrophy; those repositioning agents as orchestrators of AI workflows may find competitive advantage. The BLS itself cautioned that its 18-occupation list is neither exhaustive nor definitive, leaving room for secondary effects across adjacent roles—quality assurance, training, knowledge management—that depend on customer service volume and complexity.
U.S. starting to see heavy job losses in roles exposed to AI The Detroit News