AI’s Labor Impact: first report with anonymous data from Claude
- Dolores Abuin
- Mar 5
- 4 min read
On February 10, Anthropic released the Anthropic Economic Index (https://www.anthropic.com/news/the-anthropic-economic-index).Additionally, they published a paper (https://assets.anthropic.com/m/2e23255f1e84ca97/original/Economic_Tasks_AI_Paper.pdf) titled “Which Economic Tasks are Performed with AI? Evidence from Millions of Claude Conversations,” which you can review directly.
Anthropic is a U.S.-based artificial intelligence research and development company founded in 2021, led by former OpenAI employees, the siblings Dario and Daniela Amodei.
What is this report about?
This initiative aims to analyze the effects of AI on the labor market. The index provides data and insights based on millions of anonymized user conversations with Claude.
Claude is Anthropic’s conversational AI model, which offers a free version for users to interact with this LLM.
This report compiles data to understand how AI is being used in real work-related tasks.
Key findings
One of the most striking aspects of the study is the dataset used for the analysis.Approximately 4 million conversations with Claude were analyzed. The study focuses on AI’s impact on the workforce, using a data-driven approach to determine which tasks and occupations are adopting the technology rather than relying on assumptions.
The results show that AI is primarily used for coding and writing, which together account for nearly half of all use cases. However, AI adoption is much broader, affecting approximately 36% of occupations to varying degrees.
The report also distinguishes between two ways AI is being used:
57% use AI as an assistive tool to enhance human productivity, such as improving text or learning new skills.
43% use AI for direct automation, where tasks are executed with minimal human intervention.
Although the analysis has limitations and only reflects data from Claude's specific platform, it opens the door for other companies to adopt similar methodologies. This would allow continuous tracking of AI’s impact on work and help forecast future trends.
Anthropic's goal is to highlight the real-world use of AI in the workplace and encourage open discussions among affected sectors—including economists, lawyers, policymakers, and academics—to develop real solutions for the inevitable changes AI will bring to the job market.
What are the most revealing findings?
This initial report provides key insights into AI adoption across different sectors.It confirms that AI is currently used mainly for software development and content writing.
The first section of the report outlines how AI adoption is distributed across different jobs:
Red: Only 4% of jobs use AI intensively.
Blue: 36% of jobs use AI moderately.
Gray: 60% of jobs use little or no AI.
The second section of the report highlights how AI is more frequently used to augment human work (57%) rather than to fully automate tasks (43%).
In this sense, AI functions as a productivity-enhancing tool rather than as a massive job automation system.
However, this balance may shift rapidly as AI models continue to advance.
Anthropic’s analysis identified six professional categories where AI has a significant presence:
Computing and Mathematics (37.2%): Programming, debugging, and network troubleshooting.
Arts and Media (10.3%): Writing and content editing.
Education and Libraries (9.3%): Assisting in educational material creation.
Administration and Office Work (7.9%): Automating reports and assisting with administrative tasks.
Life Sciences (6.4%): Data analysis and scientific documentation.
Business and Finance (5.9%): Report writing and financial data analysis.
The study also shows that sectors with the lowest AI adoption are those requiring a high degree of manual or physical labor, such as agriculture, fishing, and forestry, where AI presence is almost nonexistent (0.1%).
How did they analyze millions of conversations?
To conduct this analysis, Anthropic developed a tool called Clio, which analyzes conversations with Claude while maintaining user privacy.
Clio categorized approximately one million conversations based on labor-related tasks defined in the U.S. Department of Labor’s O*NET database.This helped map the relationship between AI use and different occupations.
The findings also allowed researchers to compare AI usage frequency per occupation with its overall representation in the labor force.
For example:
Software developers accounted for 37.2% of AI usage, even though they make up only 3.4% of the U.S. workforce.
In contrast, administrative workers, who represent 12.2% of the workforce, used AI in only 7.9% of the analyzed conversations.
Which salary ranges are most affected by AI?
The study compared AI adoption rates with median wages for each occupation and found that jobs with very low and very high salaries tend to use AI the least.
The report also highlights that AI usage is most common in middle- and high-income jobs, such as programmers and data scientists, while adoption is lower in both low-wage and very high-paying professions.
This trend suggests that practical barriers and technological limitations still prevent the full automation of certain tasks.
For instance, occupations like hairstylists and obstetricians show low levels of AI adoption due to the need for manual skills or highly specialized knowledge.
Regarding the balance between automation and augmentation, the data reveals that AI collaborates with workers more often than it replaces them.
57.4% of cases involved AI assisting with validation, task iteration, and learning.
42.6% of cases involved AI autonomously completing tasks.
What's next for AI in the workplace, according to the Anthropic Economic Index?
It is still uncertain whether AI usage in these analyzed conversations directly correlates with users’ actual jobs, or whether the AI-generated results were used without modifications or human oversight.
Additionally, the study was limited to Claude.ai Free and Pro users, excluding enterprise and API-based interactions.Moreover, this analysis only focuses on the U.S. labor force.
To address these limitations and continue tracking AI’s impact on the job market, Anthropic plans to regularly update the Economic Index.
The company aims to identify long-term trends in AI automation and workplace assistance.
The Anthropic Economic Index offers a first look at AI’s impact on the labor market.Currently, AI acts more as a complement to human workers rather than replacing them.
However, this role may evolve as AI technologies advance further.
Comentarios