Anthropic Institute: inside Anthropic, junior technical hires fell from 27% to 9% as agent hours reached 6x human hours; a companion survey finds 91% of staff vs 39% of US adults expect a Nobel-level AI discovery by 2030

Why it matters

【Key points】

The Anthropic Institute published two economics reports early on Oct 17 Beijing time, treating Anthropic itself as a case study of heavy AI adoption.

"How employment is changing at Anthropic" (Wilmers, Catipon, Massenkoff, McCrory) draws on internal data and 17 interviews. The junior share of technical staff fell from 21% to 14% since early 2025, and junior share of technical hires from 27% to 9%; technical departments fell from 61% to 56% of headcount. Agent hours passed human hours in March 2026; by August, employee-directed agents logged 6x human hours. Commits per engineer are 5x six months ago; comparing 192 adopters with 74 non-adopters, agent-written code was no more likely to be deleted or reverted. Human attention is now the bottleneck: interactive sessions sit ~80% idle waiting for people, and agent-management "meta-work" rose from 10% to 22%. Anthropic says it runs an early-warning system for at-risk jobs and prioritizes redeployment.

"The AI expectations gap" (Sacher, Cotter, Charles I. Jones, Anton Korinek, McCrory) compares 10,980 US adults (Morning Consult, August) with 1,334 staff: 39% vs 91% expect a Nobel-level AI discovery by 2030; staff expect AI to do 75% of their own work by 2030 vs 28% for US workers. Their model maps the medians to 2030 GDP growth of 5.4% vs 17.1%. Staff estimate ~51% of building a frontier model is already done by AI.

【Verifiable】

Overview: https://www.anthropic.com/research/expectations-for-the-future

Report 1: https://www.anthropic.com/research/employment-at-anthropic

Report 2: https://www.anthropic.com/research/ai-expectations-gap

【Our read】

1. The junior-hiring drop is rare hard data from inside a frontier lab; watch whether it spreads.

2. All data is self-reported and unaudited; adoption isn't random, which the authors say may bias estimates.

3. Staff views are an internal poll, not company plans. Both groups reason alike about unemployment; the gap comes from lived experience.