Methodology & sources
Everything here is built from public labor data and published AI research. Data prepared July 2026.
Sources
- NYSDOL Long-Term Occupational Projections
How many people work each job today, and the 10-year job outlook—statewide and for New York's 10 regions.
- NY Occupational Employment & Wage Statistics (OEWS)
Pay for each job, statewide and by region.
- “GPTs are GPTs” (Eloundou et al., 2024)
How exposed each job's tasks are to AI.
- O*NET 30.3
Each job's tasks, work activities, skills, and related jobs.
- BLS Employment Projections: education & training
The typical education and experience needed to enter each job, plus the on-the-job training that typically follows.
- Fields-of-study–to-jobs crosswalk (NCES)
A federal table matching college and training programs to the jobs that use what they teach—so each job can show relevant fields of study. It matches program content to job skills, not where graduates actually end up.
How we put it together
- The AI-exposure measure
- We use the Eloundou et al. GPT-4 score. A task counts fully when AI alone could cut the time it takes at least in half, and at half weight when that would take extra software built around the AI. The score is national and describes what the work is like—not a prediction the job will change. We sort jobs into five bands of exposure to AI.
- Linking AI exposure to New York jobs
- The research rates jobs by O*NET code; New York's employment uses SOC codes. We match them with O*NET's official crosswalk, and where several O*NET codes map to one job we average them.
- Jobs you could switch to
- We start from O*NET's list of related jobs and look at how much their day-to-day tasks overlap with yours (weighted by how important each task is). We show the least AI-exposed moves first; jobs about as exposed as yours still appear, lower down. A recommended move must clear three bars: openings in your region, pay within about 10% of your current job or more, and never more exposure to AI. When nothing clears all three, we show the closest options anyway—labeled with exactly which bars they miss.
- Skills for a move
- For each move we show skills the two jobs share—the ones you already have—and skills the new job needs at a meaningfully higher level: the ones to build.
- How to get there
- Fields of study come from the NCES crosswalk; the preparation level—typical education, experience, and training—is O*NET's description for that job.
- Two education numbers
- “Typical entry” (BLS) is what you need to get hired. The secondary note (O*NET) is what surveyed workers say a new hire would typically need—a different measure, not the education workers themselves have.
- Outlook years don't all match
- The statewide outlook covers 2024–2034; the regional outlook covers 2022–2032. We label which one you're looking at, so statewide and regional openings aren't read as an exact comparison.
How current the data is
| Measure | Vintage | Source |
|---|---|---|
| AI exposure | 2023-era GPT-4 ratings | Eloundou et al. |
| Jobs & outlook | 2024–34 statewide · 2022–32 regional | NYSDOL |
| Wages | Adjusted to early 2026 | NY OEWS |
| Tasks, skills, related jobs | O*NET 30.3 (2026) | O*NET |
| Education | 2024–34 cycle | BLS |
| Fields of study | 2020 CIP edition | NCES |
Built by Concourse as a prototype, using public data from New York State (NYSDOL), the U.S. Bureau of Labor Statistics, O*NET, and NCES, and published research by Eloundou et al. Not an official New York State Department of Labor service.