Anthropic staff expect AI to be able to perform an average of 75% of their current work hours by 2030, compared with 28% among US workers, according to a study Anthropic published Oct. 16. The figures describe expectations about capability, not a measured share of jobs eliminated.
The study compares an August internal poll of 1,334 Anthropic staff with a survey of 10,980 US adults. The staff poll was unweighted. Morning Consult weighted the adult responses to represent the US population, though its online, opt-in panel excluded adults without internet access. For the work-hours comparison, the public sample was limited to employed respondents. Staff used a slider; the public chose ranges, which the researchers converted to midpoints.
The 17.1% GDP figure highlighted in a separate post comes from an economic model. Researchers mapped responses about capabilities, adoption, automation, productivity and worker transitions into economic scenarios. The median staff member’s answers implied annual US GDP growth of 17.1% in 2030, compared with 5.4% for the median US adult’s answers. The distributions overlapped widely.
That exercise used the 1,093 staff and 3,259 US adults who answered every required model-input question, not the full survey samples. It also held the model’s other parameters at the study’s “Substantial” scenario values. These are conditional model outputs, not observed growth or direct GDP forecasts from the respondents.
