I was demoing AI coding tools for an energy firm. One of them went quiet and asked: "What will my kids do?" I didn't have a good answer. So I went looking for one in the data.
The U.S. Bureau of Labor Statistics publishes the Occupational Outlook Handbook, covering 342 occupations and 143 million jobs across the entire U.S. economy. For each occupation, the BLS provides median pay, education requirements, number of employed workers, and projected employment growth through 2034.
In March 2026, Andrej Karpathy (co-founder of OpenAI) built an open-source tool that scraped all 342 BLS occupations and scored each one on AI exposure (0-10) using an LLM. The scoring rubric: if the job can be done entirely from a computer, exposure is high. If it requires physical presence, manual skill, or real-time human interaction, exposure is low. Karpathy published the data, the code, and the scoring prompt. He did not draw conclusions.
This analysis cross-references Karpathy's AI exposure scores with the BLS pay, education, and growth data to answer a question he left open: what do these numbers actually mean for the next generation entering the workforce?
These are the careers parents push their kids toward. Every one of them scores 7 or higher on AI exposure.
| Occupation | Education | Pay | AI Exposure |
|---|
Only 28 out of 342 occupations are growing, pay above median, and have low AI exposure. Almost all of them require being in the room.
| Occupation | Education | Pay | AI Exposure |
|---|
The Bureau of Labor Statistics is already projecting these jobs will shrink through 2034. The AI exposure scores explain why.
| Occupation | Pay | BLS Outlook | AI Exposure |
|---|
All 342 occupations. Each dot is a job. Bigger dots mean more people work there.
Search for any job or filter by category. Click a column to sort.
| Occupation | Education | Pay | AI Exposure |
|---|