Anthropic Study: AI Theory Coverage Far Exceeds Actual Adoption, White-Collar Jobs Face Unemployment Risk

Gate News reports that on March 7, Anthropic released a research report showing that although AI theoretically covers most tasks in business, finance, law, and computing, the actual adoption rate is only a small fraction. Taking the Claude model as an example, its theoretical coverage for computer and math jobs reaches 94%, but the actual usage rate is only 33%. The study introduces an “exposure” metric comparing theoretical capabilities with real-world usage data.

The research indicates that the group with the highest AI exposure is highly educated, high-income female white-collar workers: compared to lower-exposure groups, this group has 16 percentage points more women, an average income 47% higher, and nearly four times the graduate degree rate. Researchers warn that as AI capabilities improve and adoption deepens, white-collar workers may face scenarios similar to the unemployment rate doubling from 5% to 10% during the 2007-2009 financial crisis.

Currently, the impact is more on hiring slowdown rather than layoffs: in the post-ChatGPT era, the job search rate for exposed professions has decreased by 14% compared to 2022, and employment rates for young workers aged 22-25 in related fields have dropped by 16%. Some young people are choosing to pursue further education or temporarily withdraw from the labor market.