Anthropic research says AI could affect far more jobs in theory than it actually does right now. Sti

Anthropic Study: AI Could Cover 94% of Computer Jobs, but Only 33% Are Actually Using It

AI's potential is massive—but its real-world footprint? Not so much. A new report from Anthropic digs into the gap between what AI could theoretically do and what people are actually using it for. Take computer and math roles: Claude could cover 94% of the tasks. Actual usage? Just 33%.

The study introduces an "observed exposure" metric to measure this disconnect. And the group with the highest exposure? Highly educated, high-income female white-collar workers. Compared to low-exposure groups, they're 16 percentage points more likely to be women, earn 47% more, and have nearly four times the rate of postgraduate degrees.

Researchers warn that if AI adoption accelerates, white-collar workers might face a shock similar to the 2007-2009 financial crisis, when unemployment jumped from 5% to 10%. For now, the pain shows up in hiring, not firing. Job applications in exposed fields are down 14% since ChatGPT arrived. Among 22-25 year olds, employment in those areas has dropped 16%. Some are staying in school longer, others just stepping back.