Andrej Karpathy leveraged AI to assess which U.S. occupations face the highest risk from the technology, as concerns mount that a widespread job collapse could impact the broader economy.
Over the weekend, the OpenAI co-founder and former Tesla AI director shared a graphic outlining how vulnerable each occupation is to AI and automation, drawing on data from the Bureau of Labor Statistics. Each role was assigned a score ranging from 0 to 10, with 10 representing the highest level of exposure.
The overall weighted exposure score came to 4.9, and Karpathy’s data also revealed that professions with annual salaries above $100,000 had the poorest average score of 6.7, while roles paying less than $35,000 per year had the lowest exposure score at 3.4.
His chart gained rapid traction online, with many people forecasting negative outcomes for white-collar workers. Karpathy removed the data shortly afterward, however.
“This was a saturday morning 2 hour vibe coded project inspired by a book I’m reading,” he explained on X on Sunday morning. “I thought the code/data might be helpful to others to explore the BLS dataset visually, or color it in different ways or with different prompts or add their own visualizations. It’s been wildly misinterpreted (which I should have anticipated even despite the readme docs) so I took it down.”
He did not respond to inquiries about how the data was misinterpreted, or what the appropriate reading of the findings should be.
Even so, an archived version of the chart is unlikely to come as a major surprise, as it aligns with claims other analysts have made about how AI could reshape the U.S. labor market.
For instance, roles including software developers, computer programmers, database administrators, data scientists, mathematicians, financial analysts, paralegals, writers, editors, graphic designers, and market researchers all received a score of 9.
This trend comes as advanced AI tools are increasingly used to process data and generate content, completing tasks in minutes that once took knowledge workers hours, days, or even weeks to finish.
While AI is widely viewed as a productivity booster for experienced employees, mounting evidence shows companies have reduced demand for entry-level workers. More firms are also announcing layoffs and citing AI as the reason, though skeptics argue the technology is being used as a scapegoat to reverse overhiring that occurred during the pandemic.
Meanwhile, Karpathy’s chart showed that construction laborers, roofers, painters, janitors, ironworkers, and grounds maintenance workers received scores of just 1. Similarly, roles including home healthcare aides, nursing assistants, massage therapists, dental hygienists, veterinary assistants, manicurists, barbers, and bartenders got scores of 2.
Earlier this month, AI startup Anthropic released a report titled “Labor market impacts of AI: A new measure and early evidence,” which found that real-world AI adoption currently makes up only a small fraction of the tasks AI tools are technically capable of performing.
Consistent with Karpathy’s data, Anthropic’s report noted that AI could theoretically handle most tasks associated with business and finance, management, computer science, math, legal, and office administration positions. While AI adoption is still moving slowly, Anthropic stated that the workers facing the greatest risk are older, highly educated, and high-earning.
And earlier this year, a widely circulated essay from Citrini Research painted a catastrophic picture of an economy destroyed by AI, triggering a stock market selloff.
But Citadel Securities quickly debunked that doomsday scenario in a sharply critical report, pointing out that job posting data from Indeed shows demand for software engineers is actually up 11% year over year so far in 2026.
Citadel also noted that the daily use of generative AI for work remains “unexpectedly stable” and currently “presents little evidence of any imminent displacement risk.” Instead of a collapsing economy, new business formation in the U.S. is expanding rapidly, and the construction of large-scale AI data centers is currently driving a localized boom in construction hiring.
Furthermore, if automation expanded at the breakneck pace Citrini has warned of, demand for computing power would inherently rise, pushing up its marginal cost.
“If the marginal cost of compute rises above the marginal cost of human labor for certain tasks, substitution will not occur, creating a natural economic boundary,” Citadel said.