The Vanishing Jobs: Careers to Think Twice About

Two Distinct Forces Are Shrinking the Labor Market, and They Require Different Responses

Of 744 occupations in our dataset, 228 are projected to decline over the next decade. The decline is not uniform. Some careers are shrinking because artificial intelligence can perform their core tasks. Others are shrinking because the industries themselves are contracting for older economic reasons. Telling the two apart changes what students and workers should do next.

Every conversation about the future of work eventually arrives at a list of jobs that are disappearing. The list usually arrives without distinctions, treating all declining careers as a single category facing the same problem. They are not.

Cross-referencing employment projections from the Bureau of Labor Statistics with AI exposure data from the Yale Budget Lab reveals two separate patterns. There are two types of vanishing jobs, driven by different forces, affecting different workers, and calling for different responses.

Of the 228 declining occupations in our dataset, 103 face what we call AI-driven decline. They have above-median AI exposure scores, meaning their core tasks overlap closely with what AI systems can perform. The other 125 face structural decline. Their AI exposure is below median, and the forces driving them down are older: industrial automation, offshoring, shrinking industries, or shifts in consumer behavior.

The distinction shapes the advice. A worker in an AI-driven declining career has a different problem to solve than a worker in a structurally declining one. One requires adapting to new tools. The other requires finding new work entirely.

Two Types of Decline

The scatter plot below makes the divide visible. Every occupation is plotted by its AI exposure score and projected employment growth. Below the zero line, two distinct clusters appear: coral dots to the right (AI-driven decline, high exposure) and purple dots to the left (structural decline, low exposure).

Figure 1: Two types of vanishing jobs. Coral = AI-driven decline (high AI exposure). Purple = structural decline (low AI exposure, driven by automation, offshoring, or industry contraction). Interactive version available online.

The AI-driven declining careers cluster on the right side of the chart. Word processors (-36.1 percent, AI exposure +4.31), data entry keyers (-25.9 percent, +3.78), telemarketers (-22.1 percent, +4.77), and payroll clerks (-16.7 percent, +4.01) share a common task profile. Their work centers on information processing: typing, data entry, telephone communication, and rule-based calculation. These are the tasks where generative AI and automation software perform reliably well.

The structurally declining careers cluster on the left. Roof bolters (-34.2 percent, AI exposure -3.06), foundry mold makers (-25.9 percent, -3.28), hand grinders (-21.2 percent, -2.58), and forging machine operators (-18.9 percent, -2.32) share a different profile. Their work is physical, manual, and often dangerous. They are not declining because AI can replace them. They are declining because the industries that employ them have been contracting for years: coal mining, traditional manufacturing, and print production. AI has almost no foothold in these jobs because the work happens in physical environments AI cannot operate in.

The 50 Fastest-Declining Careers

The table below lists the 50 occupations with the steepest projected declines, color-coded by whether the primary driver is AI or structural forces. The mix is roughly even: 21 of the top 50 are AI-driven, and 29 are structural.

Table 1. The 50 Fastest-Declining Careers

Ranked by projected percent decline, 2024-34. Driver indicates whether decline is AI-driven (above-median exposure) or structural.

Among the AI-driven declines, the pattern is consistent across roles. Every job on this side of the list is built around processing, transmitting, and recording information in standardized ways. Word processors type formatted documents. Data entry keyers move data from one system to another. Telemarketers deliver scripted sales pitches. Payroll clerks apply fixed rules to calculate wages. The core task in each case is routine, rule-based, and well-suited to automation.

Among the structural declines, the theme is also consistent: physical work in industries that are themselves shrinking. Mining occupations dominate the top of the structural list. Manufacturing roles, forging, milling, grinding, fabricating, are declining as production moves offshore or shifts to robotics rather than AI. Print-related roles are shrinking with the publishing industry that supports them.

Where the Job Losses Are Largest in Absolute Terms

Percentage declines tell only part of the story. A career declining 30 percent from a base of 1,000 workers loses 300 jobs. A career declining 5 percent from a base of 3 million loses 150,000. For communities, educators, and workforce planners, the absolute numbers carry at least as much weight as the rates.

Table 2. The 30 Biggest Job Losses in Absolute Terms

Ranked by projected number of jobs lost (thousands), 2024-34.

Sources: BLS OEWS 2024, BLS Employment Projections 2024-34, Yale Budget Lab AI Exposure Data (Feb 2026)

Cashiers lead by a wide margin, with 313,000 projected jobs lost, more than the next two occupations combined. The role has borderline AI exposure (0.16, just above median) and is being reshaped by self-checkout technology, mobile payment systems, and automated retail. The decline is modest in percentage terms (-9.9 percent) but enormous in absolute impact because so many people currently work as cashiers.

Office and administrative roles dominate the AI-driven losses. General office clerks (-177,000), customer service representatives (-153,000), bookkeeping clerks (-94,000), and administrative assistants (-30,000) together account for roughly 454,000 vanishing positions. These are the roles where generative AI, automated document processing, and intelligent chatbots are reducing the need for human workers.

The structural losses concentrate in food service and production. Fast food cooks (-90,000), food preparation workers (-30,000), hand packers (-32,000), and various machine operators collectively represent hundreds of thousands of positions disappearing for reasons that have little to do with AI. Kitchen automation, self-service kiosks, and the steady mechanization of factory work are doing the displacing instead.

Where Jobs Are Vanishing: The Sector View

The sector breakdown shows the starkest divide. Production has the most declining occupations (67), but only 9 percent of those declines are AI-driven. The decline is almost entirely structural, rooted in manufacturing contraction. Office and Administration has 41 declining occupations, and 85 percent of those are AI-driven. This is the sector where AI is having the most direct displacement effect. Sales sits in between, with declining careers split roughly evenly between AI disruption and structural changes in retail.

What Students Should Know

For students currently considering career paths, the vanishing jobs list is less about which specific occupations to avoid and more about which types of work to be cautious about.

When the core of a job is routine information processing, typing, filing, data entry, basic bookkeeping, scripted communication, the trend is clear and accelerating. These tasks are being absorbed by software, and the careers built around them are contracting. Students drawn to office and administrative work should look for roles that emphasize coordination, judgment, and human interaction rather than processing and data management.

When the core of a job is a physical skill in a shrinking industry, the calculus changes. The work itself may not be threatened by AI, but demand for it is falling for economic and structural reasons. A student interested in manufacturing should look at the growing segments, renewable energy equipment, aerospace, medical devices, rather than traditional metalworking or mining.

Both types of decline share one common thread: specificity of skill. Workers in vanishing careers tend to have skills tied to a single type of task or a single industry. Workers who survive and thrive tend to have transferable skills, communication, problem-solving, technical adaptability, that allow them to move as the landscape shifts. The most career-resilient workers are not the ones who avoid change but the ones who build the capacity to keep moving with it.

For families weighing educational investments, the practical guidance follows from the data. Steer toward careers defined by judgment, human connection, and complex problem-solving. Steer away from careers defined by processing, repetition, and routine. The first group is growing. The second group, whether for AI reasons or structural ones, is not.

Methodology

This analysis merges BLS Occupational Employment and Wage Statistics (May 2024), BLS Employment Projections 2024-34, and AI exposure data from the Yale Budget Lab (February 2026). AI exposure is a PCA-weighted composite of six normalized metrics. Occupations are classified as “AI-driven decline” if they have above-median AI exposure (-0.29) and negative projected growth, or “structural decline” if they have below-median exposure and negative growth. Absolute job losses are calculated from BLS employment level projections (in thousands). Sector classifications use 2-digit SOC codes.