Skip to main content
← Back to publications

Evidence Brief · 劳动力转型

When Agents Become Teams

An evidence-led examination of AI agents, human oversight and early-career work at Anthropic, with a comparison to China.

Download PDF
Cover of When Agents Become Teams, AIGW Enterprise Workforce Transition Deep Dive 15.

概览

How does work change when AI agents participate in tasks alongside people? This AIGW Deep Dive examines Anthropic's published research on internal work, organizational roles and AI use, alongside independent research and a comparison with China.

The report considers how employees coordinate agents, verify outputs and retain authority, and how these practices may affect the skills and learning opportunities of early-career workers. It distinguishes reported activity and time estimates from independently verified productivity and employment outcomes.

Company-authored studies provide a window into Anthropic's own experience but do not establish economy-wide effects or a causal estimate of job losses. The China comparison illustrates related mechanisms in a different context; it is not a representative comparison of national workforces.

关键信息

  • Agent activity and human labor are different measures: estimated agent runtime does not establish equivalent human work completed or productivity gained.
  • Human oversight includes defining tasks, reviewing outputs, intervening and retaining decision authority; these responsibilities cannot be inferred from agent use alone.
  • Changes in employees’ reported work and interactive sessions describe activity within the studied setting, not a verified causal effect on overall employment.
  • Early-career pathways may change as routine tasks are automated; hiring and staffing shares alone do not establish layoffs or individual career outcomes.
  • The China comparison offers contextual evidence about alternative organizational approaches, not a representative estimate of workforce effects.

关于本出版物

Suggested citation

AIGW Policy Observatory. (2026). When Agents Become Teams: Anthropic and the Redesign of Human Work, Decision Authority and Early-Career Pathways. AIGW Enterprise Workforce Transition Deep Dive 15. https://gaeedu.org/publication/anthropic-ai-agents-human-work-supervision-entry-level-careers

References

  1. 1. Anthropic Institute. How employment is changing at Anthropic. 2026-10-16. Company-authored primary mixed-method research
  2. 2. Anthropic. How AI is transforming work at Anthropic. 2025-12-02. Company-authored antecedent survey, interviews and telemetry
  3. 3. Anthropic. Agentic coding and persistent returns to expertise. 2026-06-16. Company-authored observational coding use study
  4. 4. Anthropic. Labor market impacts of AI: A new measure and early evidence. 2026-03-05. Company-authored occupational exposure and labor-market analysis
  5. NBER / Demirer Musolff Yang. Writing Code vs. Shipping Code: Productivity Effects Across Generations of AI Coding Tools. 2026-09 revision

    5 · External working paper / matched event study

更多参考文献 (15)
  1. Stanford Digital Economy Lab. Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence. 2026-08-12 revision

    6 · External administrative payroll research / descriptive study

  2. METR. Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity. 2025-07-10

    7 · External RCT

  3. METR. We are Changing our Developer Productivity Experiment Design. 2026-02-24

    8 · External methods note

  4. Suproteem K. Sarkar / SSRN. AI Agents and Higher-Order Work. 2026-05-08 revision

    9 · External working paper abstract only

  5. 10. AIGW Policy Observatory. WTS-0043 Anthropic Signal. 2026-11 published. AIGW prior Tracker episode
  6. 11. AIGW Policy Observatory. Changan Automobile: How Domain Experts Build AI Workflows in China. 2026-10-17. AIGW company case synthesis
  7. 12. AIGW Policy Observatory. LONGi: From Quality Inspector to AI Trainer. 2026-10. AIGW company case synthesis
  8. 13. AIGW Policy Observatory. Liuzhou Steel Group: From the Shop Floor to the Control Room. 2026-10-22. AIGW company case synthesis
  9. 14. AIGW Policy Observatory. JD.com and LONGi: From Frontline Workers to AI Trainers in China. 2026-11-07. AIGW comparative case synthesis
  10. 15. State Council English portal / Xinhua. China issues guidelines to promote employment amid AI advances. 2026-10-23. Official English summary of employment-friendly directive
  11. State Council General Office (reproduced on local gov portal). 关于适应人工智能技术发展助力构建就业友好型发展方式的意见 国办发〔2026〕29号. 2026-10-23 posted; 2026-10-16 issued

    16 · Government primary policy document via official reproduction

  12. 17. AIGW Policy Observatory. Workforce Architecture for the AI Enterprise. 2026-10-12. AIGW flagship institutional framework
  13. METR. Frontier Risk Report (February to March 2026)

    18 · 2026. Independent safety research context

  14. 19. MOHRSS NDRC MIIT NDA. 关于加快推进“人工智能+人社”应用发展的实施意见. 2026-07-08 official press notice. China official policy notice
  15. 20. MIIT and seven ministries. “人工智能+制造”专项行动实施意见. 2026-01-07 published (issued 2025-12-25). China official policy