A recent survey conducted by Anthropic in August 2026 reveals significant differences in expectations regarding artificial intelligence (AI) capabilities between U.S. adults and Anthropic staff. While 39% of U.S. adults believe AI could achieve a Nobel-level discovery by 2030, a striking 91% of Anthropic employees hold the same belief. This divide illustrates broader contrasting views about the future of AI, especially regarding productivity gains and job automation.
Read More
Anthropic staff predict more substantial productivity improvements from AI, with 59% stating that tasks would take a quarter of the time or less when assisted by AI, compared to 26% of U.S. adults who feel the same. Regarding their own jobs, over half of Anthropic staff anticipate AI will be able to perform at least 80% of their tasks by 2030, while only 5% of U.S. workers share this expectation.
Both groups, however, align in their views on the potential for higher unemployment. Those believing AI will take over more of their current work anticipate elevated unemployment rates. Despite their differing perspectives on AI's capabilities, there is agreement on the broader potential economic impact.
The study also examined how both groups view AI's capabilities and adoption hurdles. While both agree on current AI functionalities, they are divided on the speed of future improvements. Among those using AI at work regularly, approximately 56% believe AI will be capable of performing at least seven out of eight analyzed tasks by 2030, contrasting with just 41% of the general public.
When exploring the barriers to AI's workplace adoption, U.S. adults specifically highlighted customer preferences for human interaction, data privacy concerns, and reliability issues as significant obstacles, whereas Anthropic staff did not identify these issues as barriers to the same extent.
Differing views on AI's role in the workplace also emerged, with Anthropic staff forecasting AI completing tasks autonomously or in tandem with humans. Conversely, 40% of U.S. adults expect that humans will primarily manage these tasks themselves, indicating more skepticism about AI's autonomy.
Lastly, expectations differ around how quickly displaced workers can find new roles in a changing labor market. Anthropic staff believe it will take an average of 12 months for a displaced worker to transition to a new job, compared to a median estimate of 7-12 months among U.S. adults.
Ultimately, the survey highlights a gap in perspectives on AI’s transformative potential, underscoring varying confidence levels between industry insiders and the general public regarding AI's future impact on work and the economy.