Anthropic’s study, published October 16, 2026, found that employees expected AI could perform an average of 75% of their current work by 2030, compared with 28% among employed U.S. respondents. Those are survey expectations about future work, not measured automation. Anthropic’s report, The AI expectations gap, also compares respondents’ views on AI’s capabilities, productivity and economic effects.
How the surveys were conducted
Anthropic’s unweighted internal poll received responses from 1,334 employees between August 9 and 12, 2026. Morning Consult surveyed 10,980 U.S. adults online between August 11 and 23. Its respondents came from an opt-in panel, and their answers were weighted to represent U.S. adults; Anthropic noted that weighting may not correct every difference between the panel and the broader population.
The response formats differed, too. Anthropic employees used a slider to estimate AI’s share of their work, while U.S. respondents chose ranges that Anthropic converted to midpoints. The results compare two groups’ expectations, but they do not come from identical samples or survey methods.
What respondents expected AI to do
More than half of Anthropic employees expected AI to perform at least 80% of their work by 2030; 5% of U.S. workers expected the same. Other answers also showed a gap between staff and U.S. respondents:
| Survey measure | Anthropic employees | U.S. respondents | Scope |
| Expect AI to make a Nobel-level discovery by 2030 | 91% | 39% | U.S. result is from the August 2026 survey. |
| Say an AI-suited task takes a quarter of the time or less with AI | 59% | 26% | U.S. figure covers adults who use AI at work; the question counted time spent checking and fixing AI output. |
Who the unemployment figure describes
The study’s 12.2% mean unemployment expectation for 2030 applies to U.S. workers who expected AI to perform at least 80% of their own work. It is a subgroup result. Anthropic reported that employee and U.S. worker expectations about unemployment closely tracked one another at similar levels of expected exposure to AI in respondents’ own jobs.
How the GDP estimates were generated
The study’s median estimates implied GDP growth by 2030 of 17.1% from employee responses and 5.4% from U.S. adult responses. These figures came from a model using complete answers from 1,093 Anthropic employees and 3,259 U.S. adults. It combined responses about AI capabilities, adoption, automation, productivity and obstacles to workers changing occupations, while holding other parameters at the values in the “Substantial” scenario.
The GDP percentages are model-implied results shaped by those inputs and assumptions, rather than measurements of economic growth.
