Skip to content

Federal data × LLM scoring

AI Career Stats

About the Project

About AI Career Stats

Bridging the gap between official U.S. government labor statistics and rapid Generative AI capability changes.

Our Mission

Discussions around artificial intelligence in the workplace often swing between alarming doomsday headlines and dismissive hand-waving. Workers, career switchers, and students are left wondering: Which specific parts of my job can AI actually do, and what human strengths remain durable?

The federal government’s standard labor sources — the U.S. Department of Labor’s O*NET program and the Bureau of Labor Statistics (BLS) — track wages, typical education, and 10-year employment growth. However, official government statistical cycles move slowly, and federal tables do not publish Generative AI automation vulnerability metrics.

AI Career Stats was created to fill that gap. We synthesize standardized O*NET 30.3 task statements with state-of-the-art LLM capability evaluation to produce granular, task-by-task automation scores — weighted by real-world job importance and grounded in official BLS wage and growth data.

Our Editorial & Modeling Principles

Task-Level Granularity

Jobs are not monoliths. A software developer doesn’t just write boilerplate code; they negotiate requirements and debug distributed systems. We evaluate specific O*NET tasks rather than painting entire occupations with a broad brush.

Importance-Weighted Math

Incidental administrative tasks should not skew a doctor's or surgeon's score. We weight each evaluated task by its official O*NET importance rating so core professional responsibilities dominate the index.

Durable Work Highlight

We explicitly identify safe and durable human tasks — physical dexterity, clinical empathy, ethical discretion, and complex interpersonal navigation — to guide workers toward career durability.

Objective Labor Context

We pair AI risk with official BLS median wages and 2025–2035 employment change projections to show real economic realities, not theoretical speculation.

How We Generate Data

Our data pipeline extracts Core work tasks and Detailed Work Activities (DWAs) from O*NET Database 30.3. Each task is evaluated against current Generative AI capabilities using an automated evaluation prompt. Scores range from 0 (completely insulated / physical) to 100 (fully automated by text/code/vision synthesis).

Read the complete mathematical formula and dataset specifications on our Methodology page.

Get in Touch

Contact & Editorial Inquiries

We value feedback from domain practitioners, economists, educators, and working professionals.

Data Corrections & Feedback

Notice an O*NET task weighting or SOC code mismatch? Send your feedback with the occupation title and URL.

corrections@ai-career-stats.com

General & Press Inquiries

For media questions, research citations, or general platform inquiries:

contact@ai-career-stats.com

Mailing Address: AI Career Stats, 1209 Orange St, Wilmington, DE 19801, USA

Response time: We aim to respond to all editorial and research inquiries within 2 business days.