Anthropic's expectations gap study, published by the Anthropic Institute on October 16, 2026, put the same questions to 10,980 US adults and to Anthropic's own staff. The two groups largely agree on what AI can do today, but diverge sharply on how far it will go by 2030. On a Nobel-level discovery, 91% of staff expect AI to achieve one before 2030, compared with 39% of the public, and a sizable share of adults answered that it will never happen.
The widest gap is about people's own jobs
Asked how much of their own work AI could do by 2030, staff averaged 75%, while US workers averaged 28%. More than half of staff expect AI to handle at least 80% of their work; only 5% of working adults say the same. Reported time savings differ too: 59% of staff say a well-suited task takes a quarter of the time or less with AI, versus 26% of adults who use AI at work, 29% of whom say it saves no time at all or takes longer.
Run through Anthropic's economic scenarios model, the median answers imply 2030 US GDP growth of 17.1% for staff and 5.4% for the public. The disagreement is not mainly about how unemployment works: in both groups, the more of their own work people expect AI to absorb, the higher their 2030 unemployment forecast, and those expecting AI to take 80% or more of their work forecast an average unemployment rate of 12.2%.
Staff are not simply optimists
Even while expecting most of their tasks to be automated, 95% of staff expect Anthropic's headcount to keep growing over the next year, with a median expectation of 66% growth; for them, automating tasks and shedding jobs are different forecasts. Staff are, however, more pessimistic about re-employment, estimating that a worker displaced by AI in 2030 who must switch occupations would need about 12 months to find a new role.
The obstacles the public ranks highest are also telling: customers preferring to deal with a person comes first, ahead of data privacy and security, reliability, and laws and liability. What slows AI diffusion over the next few years may be trust and accountability rather than raw capability. As our earlier report on Anthropic's internal agents already outnumbering employees six to one showed, staff extrapolate from tools they already rely on daily. Read as a diffusion map, the survey suggests what happens first in frontier-adjacent jobs tends to reach other industries a few years later, in diluted form.
