Anthropic maps which jobs robots can do
Anthropic's new robot exposure index finds that robots can already perform 74% of physical US job tasks, yet they are cost-competitive with human workers for just 0.3% of them.
Anthropic has published a robot exposure index that measures how well present-day robots can perform US job tasks. It follows the company's earlier research on how exposed jobs are to large language models.
The researchers used Claude to score roughly 19,000 task descriptions from O*NET, a database covering about 900 occupations. They defined robots as "autonomous physical machines that sense and act," and ranked each task by the kind of work environment a robot needs to perform it. The tiers run from E0, where a robot cannot do the task, to settings that are less and less controlled.
Headline findings:
- Robots can perform 74% of physical tasks, which account for 34% of US working hours. Most of that capability is limited to structured environments.
- Combined with LLMs, robots expose about 80% of job tasks by working time. The work left over is highly interpersonal or requires physical skills robots don't yet have.
- Robots are cost-competitive with human labor for only 0.3% of tasks. If robot prices keep falling at their historical rate, that share would take about 40 years to reach 10%.
- Driving and warehouse jobs are highly exposed. Nursing and general repair are not. Exposed workers are more likely to be male, less educated and lower paid.
The authors ran a backtest from 1977 to today. Over that period, jobs more exposed to existing robots saw larger declines in wages and employment. Robots also became able to do about 2% of the physical work they previously couldn't each year.
After cost, the report names capability as the main barrier, such as the dexterity needed to untangle wires. Human preferences and regulation also limit adoption. One caveat is that the exposure scores come from Claude's own assessments.