Anthropic's "AI Expectations Gap" Survey: 91% of Staff vs 39% of US Adults Expect Nobel-Level AI by 2030

Anthropic's Institute published "The AI Expectations Gap," a survey of 10,980 US adults and Anthropic staff conducted in August 2026, finding that the two groups largely agree on what AI can do today but diverge sharply on how fast it will improve, how widely it will be adopted, and how much time it will save. The gap matters because, run through Anthropic's economic model, the median US adult's answers imply 5.4% GDP growth in 2030 while the median staff member's answers imply 17.1%.

The report is authored by Szymon Sacher, Tess Cotter, Charles I. Jones, Anton Korinek, and Peter McCrory, and published as a technical report by The Anthropic Institute. The US adult sample is weighted to be nationally representative; Anthropic staff responses are unweighted and described as personal views gathered in an internal poll, not company forecasts.

Key findings at a glance

• 39% of US adults and 91% of Anthropic staff expect AI will be able to make a Nobel-level discovery by 2030.
• Among US adults who use AI at work, 26% say a task takes a quarter of the time or less with AI, compared with 59% of Anthropic staff.
• Over half of Anthropic staff expect AI to be able to do at least 80% of their current work by 2030, compared with 5% of US workers.
• In both groups, people who expect AI to do more of their own work anticipate higher unemployment, and at the same level of expected automation both groups forecast roughly the same unemployment rate.

The two groups agree on today's capabilities but not on the pace of progress

US adults and Anthropic staff rank eight tasks — from writing routine emails to making a Nobel-level scientific discovery — in nearly the same order of difficulty, and they largely agree on what AI can do today, particularly for complex tasks.

The disagreement is about the future. Around a quarter of both groups think AI can design, conduct, and publish scientific studies today. By next year, 9 in 10 staff expect this to be feasible, compared with 38% of the general public. By 2030, 39% of US adults expect AI to be able to make a Nobel-level discovery, compared with 91% of staff.

The sharpest divergence is in the tail of the distribution: depending on the task, between 19% and 40% of US adults say AI will never be able to do it, compared with at most 2% of staff.

Frequency of AI use explains part of the difference. About a fifth of respondents use AI at work regularly; 56% of that group expect AI to be able to do at least seven of the eight named tasks by 2030, versus 41% of all US adults and 28% of those who have never used an AI tool. Staff sit at 96%.

Adoption is expected to lag capability, and the public blames people and rules more than technology

Capability alone does not determine economic impact, because people and businesses must actually use AI. Of the work tasks AI will be able to do, US adults expect 42% will be done with AI by 2030; Anthropic staff expect 65%.

Asked to pick up to three of nine obstacles to AI use at work in 2030, US adults ranked "customers preferring to deal with a person" first, followed by "data privacy and security concerns," "AI not being reliable enough," and "laws, regulations or liability." The report notes that the general public expects people's preferences and rules, more than technological constraints, to hold adoption back.

Regular AI users at work rank the obstacles differently: 42% of them put data privacy and security concerns in their top three, compared with 37% of the rest, while 37% cite customer preferences for humans, compared with 42% of the rest.

Anthropic staff expect far more autonomy and far larger time savings

On how AI and humans will divide work by 2030, the vast majority of Anthropic staff say AI will either complete tasks alone or with a worker evaluating the work. By contrast, 4 in 10 US adults expect humans will do most of the work themselves or with some AI help.

On time savings, 85% of Anthropic staff say a task for which AI is well suited takes half the time or less. Among US adults who use AI at work, the typical answer is half the time, and 29% say the task takes just as long or longer, versus 5% of staff. A quarter of the time or less is reported by 26% of US adults who use AI at work, 21% of those who do not use AI at work, and 59% of staff.

The report connects staff answers to observed internal outcomes: Anthropic employees' lines of code added tripled in the month after adopting coding agents, and sales representatives closed more deals with clients after adopting AI agents, with output per person also rising in functions such as legal and recruiting.

Staff expect AI to do most of their own work — but not to eliminate their jobs

On average, Anthropic staff say AI will be able to do 75% of their current work hours by 2030. US workers put the share at 28% on average, and only 1 in 20 say AI could do 80% or more of their work hours by 2030.

Expectations vary by occupation among US workers: 25% of work hours for manual trades, production, and transport workers, and 33% for workers in computer, engineering, and science occupations. Within that last group, regular AI users expect only 40% — closer to the average US worker than to the average Anthropic staff member.

Despite that, 95% of Anthropic staff expect Anthropic's headcount to grow over the next 12 months, the median staff member expects it to grow by 66%, and 27% expect it at least to double. The report states these are personal expectations from an internal poll and do not represent Anthropic's plans or its view of its headcount, adding that "automating tasks and shedding jobs are different forecasts, at least at a frontier AI lab."

Own-job exposure predicts unemployment forecasts in both groups

Respondents were told the current US unemployment rate is about 4.2% and asked to forecast the 2030 rate. US workers who expect AI to do less than half of their work forecast unemployment of 7% to 8%. Above half, the forecast rises with expected exposure: those who expect AI to do 80% or more of their work forecast 12.2% unemployment on average, above the peak during the Great Financial Crisis.

Across US workers, ten more percentage points of work affected is associated with 0.4 points more forecasted unemployment. The report states that general optimism about AI does not explain this: holding own-work exposure fixed, the share of the eight tasks a worker expects AI to master is unrelated to the forecast.

The same gradient appears among Anthropic staff, whose forecasts closely mirror those of the general population despite much higher reported exposure. The report concludes that the two groups do not hold different theories of how AI affects their own jobs and the broader labor market — they have different experiences of what AI capabilities are, and therefore different expectations.

On re-employment, Anthropic staff are slightly more pessimistic than US adults. The median staff member expects a worker displaced by AI automation in 2030 who needs to switch occupations to take 12 months to find a new role, compared with a median US adult response in the range of 7–12 months.

Model-implied economic scenarios: 5.4% versus 17.1% GDP growth in 2030

Using the framework from Economic Scenarios for Transformative AI, the authors translate each respondent's answers across five dimensions — capabilities, diffusion, automation, productivity, and worker transition frictions — into model-implied economic outcomes. They identify every respondent who answered all key model inputs: 3,259 US adults and 1,093 staff.

The median US adult's answers imply GDP growth of 5.4% in 2030. The median staff member's answers imply GDP growth accelerating to 17.1% by 2030. The report notes significant variation within each group, so the distributions overlap widely.

An interactive scenario explorer published at anthropic.com/institute/econ-scenarios on September 9, 2026 drew 23,075 visitors who completed a shorter-form version of the survey in the first 48 hours, connecting from more than 140 countries. Their answers place them between US adults and Anthropic staff, with the median visitor's answers implying 12.5% growth. The report cautions that these visitors chose to visit an Anthropic page and take the quiz, so they are likely more interested in AI than the general public.

The model's three named scenarios are Modest (2.4% annual GDP growth), Substantial (5.4%), and Extreme (15.4%). The report's conclusion states that staff expectations imply "by 2030, GDP is doubling in size roughly every four and a half years."

Inside the lab: staff estimate AI already does half the work of building a frontier model

Asked what share of the work of building an early-2026 frontier model is done by AI today, the median Anthropic staff answer is 51%, rising to 80% by the end of 2027 and 95% by the end of 2030.

Estimates vary far more within teams than between them. Staff on research teams, who are closest to model building, put AI's share today somewhat higher at 60% on average, against 53% for all other survey respondents. The report says this is consistent with information playing some role, while noting that researchers may also use AI more in their own work by virtue of what they do.

These responses, like the hiring forecasts, are personal views from an internal poll fielded to Anthropic staff in August 2026 and are not a measurement by Anthropic or a company forecast. The report references a separate Anthropic Institute report on progress toward recursive self-improvement (Favaro and Clark, 2026).

Why the gap may narrow

The report's central interpretation is that day-to-day experience with AI differs widely between the two groups. Most Anthropic staff already work alongside AI that does much of their job, whereas few US workers do. Those who do already hold expectations closer to those of Anthropic staff, which the authors say suggests public expectations may shift as AI use spreads — with the caveat that how far they shift depends on whether AI comes to do as much of the work elsewhere as it does at a frontier lab.

Methodology and caveats

The survey was fielded in August 2026 to a nationally representative sample of US adults and to Anthropic staff. US adults' responses (N=10,980) are weighted to be nationally representative; Anthropic staff responses are unweighted. Staff answered the own-work-hours question on a slider while US adults chose a range, with range midpoints used in the analysis. The unemployment question told respondents the current rate was about 4.2%. The obstacle question was not asked of Anthropic staff. The report's suggested citation is Sacher, Cotter, Jones, Korinek, and McCrory (2026), "The AI Expectations Gap: Public and Frontier Lab Views on AI's Impact on Work," The Anthropic Institute, Technical Report.