What the Numbers Actually Show
Every week brings a new headline about AI eliminating jobs or creating them. The reality is messier, more interesting, and more navigable than either narrative suggests. This post pulls together data from the World Economic Forum, McKinsey Global Institute, LinkedIn's Economic Graph, and the U.S. Bureau of Labor Statistics to give a clearer picture of where things stand in 2026.
72%
of Fortune 500 companies now use AI in at least one hiring stage (LinkedIn Talent Trends Report, 2025)
That 72% figure is striking, but it needs context. "Using AI in hiring" covers an enormous range, from a basic ATS that parses resumes to fully automated async interview platforms that score candidate responses. The 72% includes companies doing either. Perhaps 18% have AI meaningfully integrated into decision-making rather than just administrative tasks.
Which Roles AI Is Actually Replacing
The jobs most at risk share common characteristics: high repetition, well-defined rules, limited physical interaction, and outputs that can be easily verified. The WEF's Future of Jobs Report 2025 identified the following categories as seeing the sharpest headcount reductions attributable to automation.
| Role Category | Estimated Reduction by 2028 | Primary Driver |
|---|---|---|
| Data entry clerks | 26% | Document AI, LLM-based extraction |
| Bank tellers | 22% | Self-service apps, AI chatbots |
| Paralegal / legal research | 17% | LLM document review |
| Customer support tier 1 | 31% | Conversational AI, self-service |
| Basic content writing | 19% | Generative AI |
| Radiological image review (assist) | 14% | Medical imaging AI |
Notice that none of these are complete eliminations. Most represent a reduction in the number of people needed for a given volume of work, not the disappearance of the role entirely. A law firm that needed 12 paralegals for document review may need 7. The work that remains tends to be more complex and higher-value.
Which Roles AI Is Augmenting
Augmentation is a quieter story than replacement, but it's where most of the actual employment impact is happening. Roles being augmented see productivity increases that allow companies to either serve more customers with the same headcount, or reduce headcount while maintaining output. McKinsey estimates that 60% of workers in augmented roles have seen their individual output capacity increase by 20-40% since 2023.
Software engineers using AI coding assistants are completing features roughly 35% faster, per a 2025 GitHub study of 50,000 developers. That doesn't mean 35% fewer engineers. It means teams can ship more product with the same team, which often means more engineers, not fewer, as companies compete to ship faster.
Where New Jobs Are Being Created
4.1M
new jobs in AI-adjacent fields projected to open in the U.S. alone by 2027 (World Economic Forum, Future of Jobs 2025)
The roles seeing the strongest demand growth are concentrated in a few areas. AI infrastructure roles, including ML engineers, AI safety researchers, and model operations specialists, have seen posted salaries jump 34% since 2023. Demand outpaces supply by roughly 3 to 1 in most U.S. metro areas.
Healthcare technology is a second growth area. Clinical informatics, health data analysts, and patient-facing AI system managers are all seeing above-average demand. The BLS projects healthcare occupations overall to grow 13% through 2031, with tech-integrated roles growing at roughly double that rate.
Infrastructure and skilled trades are also rebounding. AI can't lay fiber cable, install EV charging infrastructure, or retrofit HVAC systems. These roles are seeing strong demand from the Inflation Reduction Act's ongoing capital deployment and the data center construction boom driven by AI compute demand.
Salary Trends in AI-Adjacent Roles
Median Salary Growth 2023 to 2026 by Role Category
What This Means for Job Seekers and Hiring Teams
For job seekers, the pattern is clear: roles that sit at the intersection of domain expertise and AI tooling are commanding the largest premiums. A nurse who can interpret AI diagnostic output and communicate it to patients is more valuable than one who cannot. A marketer who can build and manage AI content workflows at scale is more valuable than one producing content manually.
For hiring teams, the implication is a candidate supply shortage in the fastest-growing roles. ML engineers and AI product managers are not abundant, and the companies that have moved to structured, efficient hiring processes are winning those candidates before others even get to the first round.
The broader picture is one of disruption concentrated in specific pockets, not a wave that drowns all employment. Total U.S. employment was at record highs through 2025. The unemployment rate for college graduates stayed below 3%. The pain is real for specific roles in specific industries. The overall labor market is not in crisis.