Anthropic Report Reveals Wide Gap in AI Economic Expectations Between General Public and Frontier Developers

Anthropic released a report titled "The AI expectations gap" on October 16, 2026, analyzing the divergence in economic outlooks between the US general public and frontier AI developers. While both groups agree on current AI capabilities, Anthropic employees anticipate significantly faster progress and greater economic disruption than the US adult population.

Divergent Views on Capability and Productivity

The study found that 91% of Anthropic staff expect AI to make Nobel Prize-level discoveries by 2030, compared to only 39% of US adults. Similarly, 90% of staff believe AI will design, conduct, and publish scientific research by 2027, whereas only 38% of the public agrees. In terms of productivity, 59% of staff expect AI to reduce task time to less than a quarter of the original duration, compared to 26% of US adults who use AI. Anthropic’s internal data supports this, showing engineers added three times more code after adopting coding agents.

Labor Market and Unemployment Projections

There is a stark contrast in how each group views the impact of AI on their own jobs. A majority of Anthropic employees believe AI could perform over 80% of their current work hours by 2030, whereas only 5% of US workers share this view. On average, staff expect AI to handle 75% of their work time, compared to 28% for US workers. Despite these differences in exposure, both groups exhibit a similar correlation between AI exposure and unemployment fears: for every 10 percentage point increase in AI-exposed work, US workers project a 0.4 percentage point rise in unemployment. Workers expecting high AI displacement predict a 2030 unemployment rate of 12.2%, significantly higher than the current 4.2%.

Economic Scenarios and GDP Growth

Using the Economic Scenarios for Transformative AI model, the report estimates 2030 GDP growth based on survey responses. The median response from US adults suggests 5.4% annual GDP growth, while Anthropic employees’ median responses imply 17.1%. Visitors to Anthropic’s online scenario explorer, who are presumed to be more AI-engaged, projected a median growth of 12.5%. The report notes that while there is consensus on current AI capabilities, the gap in expectations regarding future speed, adoption, and productivity gains remains wide. However, US adults who regularly use AI at work hold expectations closer to those of Anthropic staff, suggesting public perception may shift as AI adoption increases.