Anthropic found in a survey study of 9,700 Claude users that so-called automation users, who hand over entire tasks to AI, showed positive expectations across all six job indicators including pay, job security and chances of re-employment.

Anthropic on June 26 (local time) released the "Anthropic Economic Index Cadences" report. It analysed usage patterns by randomly sampling responses to the Anthropic Economic Index Survey launched in April and actual Claude usage data in a de-identified way.

The report said users with higher automation use rates were positive about AI's impact not only on economic indicators such as pay, job security and re-employment, but also on intrinsic indicators such as the meaning of work, autonomy and human interaction. The largest expectation gaps were in pay increases and chances of re-employment.

Perceived productivity was also high. Some 86 percent of respondents said work became faster. A majority also responded positively on work scope (82 percent) and quality (69 percent). Some 68 percent said they learned more because of AI, and 57 percent said they felt their skills had gained market value.

However, 10 percent of respondents said the likelihood that they would lose their jobs was high or very high. In particular, a larger share worried about junior colleagues' jobs disappearing than worried about their own jobs. More than a third also said their job responsibilities would change significantly within 12 months.

Perceptions of AI's ability to replace work varied by country and career stage. Respondents in lower-income countries and those with shorter careers were more likely to say AI could perform more work. By contrast, expectations that AI capabilities would expand after 12 months were evenly high across respondents regardless of country and career stage. More than a third of respondents expected that AI would be able to handle most of their work within the next year.

Usage patterns showed a clear gender gap. Female users had an automation rate 7.3 percentage points lower than men and a Claude Code usage rate 6.3 percentage points lower. Instead, they logged more active time in chat, showing a tendency to use AI in repetitive and collaborative ways.

The report also analysed how adoption is spreading across industries. It said that while most early Claude users were in high-tech occupations, recent users are using Claude for relatively lower-paid work.

This analysis covered user data for Claude, Claude Cowork and Claude Code.