# anthropic-ai-transforming-work-research-2025-12-02

## Veille

Anthropic Research - AI Work Transformation - Claude Code Impact - Software Engineering - AI Adoption - Productivity Study - Workplace Evolution - AI Collaboration - Skills Development - Future of Work

## Titre Article

How AI is transforming work at Anthropic

## Date

2025-12-02

## URL

https://www.anthropic.com/research/how-ai-is-transforming-work-at-anthropic

## Keywords

Anthropic, AI Transformation, Workplace Impact, Claude Code, Software Engineering, AI Adoption, Productivity Study, Workplace Evolution, AI Collaboration, Skills Development, Future of Work, AI-Augmented Workplace, Engineering Productivity, AI Delegation, Technical Expertise, Workplace Dynamics, Career Evolution, AI Economic Impact

## Authors

Anthropic Research Team (132 engineers and researchers surveyed, 53 in-depth interviews conducted)

## Ton

**Profil:** Institutional research | Collective team voice | Empirical-Nuanced | Tech professionals and decision-makers

An Anthropic research publication with an institutional, empirical tone: explicit methodology (survey of 132 engineers, 53 interviews, internal usage data), quantified results, and direct employee quotes. The register is balanced: productivity gains are presented alongside concerns (skill atrophy, isolation, career uncertainty), without triumphalism. Authority comes from the position of privileged observer ("early adopters" within the company developing the AI). Target audience: engineers, tech leaders, and policymakers anticipating workplace transformations.

## Pense-betes

- **Methodology**: survey of 132 Anthropic engineers/researchers, 53 in-depth interviews, analysis of internal Claude Code usage data (August 2025, Claude Sonnet 4 and Opus 4 models)
- **Dominant use**: debugging, code comprehension, error correction, learning the codebase
- **Key figures**: Claude used for 60% of work; ~50% productivity gain (2-3x vs previous year); 27% of assisted work would not have been done otherwise
- **Limited delegation**: only 0-20% of work can be "fully delegated"; active supervision required for critical work
- **Delegation intuition**: verifiable, low-stakes, or tedious tasks first; design and "taste" remain human but the boundary is shifting
- **Expansion vs. atrophy**: developers become "full-stack" outside their expertise, but fear of atrophy of deep skills ("when producing is so easy, it becomes difficult to take the time to learn")
- **Social dynamics**: Claude replaces colleagues as the first technical resort; less mentorship; juniors ask seniors fewer questions
- **Growing autonomy of Claude Code**: from 10 to 20 autonomous actions in six months; code design 1% → 10%; new features 14% → 37%
- **"Papercuts"**: 8.6% of tasks = fixes for minor issues typically deprioritized
- **Career**: short-term optimism, long-term concern ("AI will eventually do everything and make me obsolete")
- **Internal responses**: new mentorship models, programs to maintain deep skills, reference to Anthropic's blog on economic policy

## RésuméDe400mots

This study by Anthropic examines how AI, and more specifically Claude Code, is transforming the work of its own software engineers. It draws on a survey of 132 engineers and researchers, 53 qualitative interviews, and analysis of internal usage data (August 2025, Claude Sonnet 4 and Opus 4 models). Its position is distinctive: observing "early adopters" within the very company developing the AI, likely representative of transformations to come elsewhere.

The figures are striking. Employees report using Claude for 60% of their work and estimate a productivity gain of about 50%, 2 to 3 times higher than the previous year. Above all, 27% of Claude-assisted work simply would not have been done otherwise: more ambitious projects, "nice-to-have" dashboards, exploratory work that is not profitable to do manually. Delegation nonetheless remains supervised: only 0 to 20% of work can be "fully delegated," with human validation still required for critical work.

The interviews reveal deep qualitative transformations. Engineers develop an intuition for delegation: first handing off verifiable, low-stakes, or tedious tasks, then progressively expanding — design and "taste" remaining human, for now. Developers become "full-stack," working competently on domains they would not previously have dared to touch. But a paradoxical concern emerges: the atrophy of the deep skills needed to write and critique code — "when producing results is so easy and fast, it becomes difficult to really take the time to learn."

Social dynamics are also evolving: Claude replaces colleagues as the first port of call for technical questions, reducing opportunities for mentorship. One senior engineer says with sadness that juniors no longer come to see him. On the career front, feelings are contradictory: short-term optimism (higher-level work, managing AI systems), long-term concern ("AI will eventually do everything and make me obsolete").

Usage data confirms growing autonomy: from 10 to 20 autonomous actions in six months, code design rising from 1% to 10% of uses, implementation of new features from 14% to 37%. 8.6% of tasks involve fixing "papercuts" previously deprioritized.

Anthropic draws internal initiatives from this — new mentorship models, maintaining deep skills — and stresses the importance of actively preparing the transition to an AI-augmented workplace, lessons transferable to other organizations.

## GrapheDeConnaissance

- Anthropic —publie→ étude sur la transformation du travail (DOCUMENT, 0.98)
- Claude Code —améliore→ productivité ingénieurs logiciels (CONCEPT, 0.95)
- Anthropic —mesure→ 132 ingénieurs et chercheurs enquêtés, 53 entretiens (MESURE, 0.97)
- Claude Code —s_applique_à→ 60% du travail des employés (CONCEPT, 0.9)
- Anthropic —mesure→ 50% de gain de productivité grâce à l'IA (MESURE, 0.92)
- Claude Code —permet→ travail supplémentaire (27%) (CONCEPT, 0.9)
- IA —réduit→ interactions de mentorat (CONCEPT, 0.85)
- IA —permet→ transition des ingénieurs vers la gestion de systèmes IA (CONCEPT, 0.88)
- Claude Code —utilise→ Claude Sonnet 4 (TECHNOLOGIE, 0.95)
- Claude Code —utilise→ Claude Opus 4 (TECHNOLOGIE, 0.95)
- IA —réduit→ compétences techniques profondes (risque d'atrophie) (CONCEPT, 0.85)
- Claude Code —permet→ développeurs full-stack (CONCEPT, 0.88)
- Anthropic —mesure→ actions autonomes Claude Code passées de 10 à 20 en six mois (MESURE, 0.9)

---
Canonical: https://www.thekb.eu/en/fiches/anthropic-ai-transforming-work-research-2025-12-02/