Andrei Karpathy introduced a new project called karpathy/jobs. He gathered data on 342 professions from the BLS statistics—approximately 143 million workers in the USA—and used large language models to assess how susceptible each profession is to automation with artificial intelligence. He assigned scores on a scale from 0 to 10.
He displayed the results in a clear diagram—a treemap—that shows the distribution of automation risks across different jobs. On average, the scores across the entire spectrum of work were around 5.3 out of 10.
Here are some examples:
– Software developers — score 8–9
– Roofers — only 0–1
– Medical data analysts — maximum score of 10 out of 10 💀💀
The basic principle is quite simple: if all activities are performed at a computer, the risk of automation and machine replacement is significantly higher. Conversely, if the work involves physical labor or unpredictable conditions, the likelihood of job loss due to machines is minimal.
According to Karpathy, nearly 57 million workers in the USA—about 40% of the entire workforce—are at high risk of change due to AI influence.
More details can be found here: https://karpathy.ai/jobs/
Created with n8n:
https://cutt.ly/n8n
Created with syllaby:
https://cutt.ly/syllaby