Anthropic Staff See AI Doing 75% of Their Work by 2030

Anthropic published a report titled "The AI Expectations Gap" on October 16, 2026, comparing its own employees' forecasts about artificial intelligence with those of US adults. The report draws on responses from 1,334 Anthropic employees and a survey of about 11,000 US adults, according to aitimes.com.

The gap is widest on science. 91% of Anthropic employees said AI could make a Nobel Prize-level scientific discovery by 2030, compared with 39% of the US general public. 90% of employees expected AI to design and carry out scientific research and publish papers by 2027, against 38% of the public.

Employees also see automation reaching further into their own jobs. They estimated AI could handle an average of 75% of their current work hours by 2030, while US workers put the figure at 28%. More than half of Anthropic employees said AI could perform over 80% of their work, compared with 5% of general workers.

59% of Anthropic employees said AI could cut the time needed for suitable tasks to a quarter or less, versus 26% of US workers who use AI at work. US tech and science workers who regularly use AI expected the technology to handle an average of 40% of their work by 2030.

Jobs and growth forecasts

Expectations tracked with views on unemployment. Workers who expect AI to do under half their job forecast average 2030 unemployment of 7-8%, while those expecting over 80% forecast 12.2%. The report's economic model put median 2030 GDP growth estimates at 5.4% for the US public and 17.1% for Anthropic employees.

Anthropic chief economist Peter McCrory said the company adopts AI about a year and a half ahead of the broader economy, making its internal changes a leading indicator.