# Working with AI: Measuring the Occupational Implications of Generative AI

**Authors:** Kiran Tomlinson, Sonia Jaffe, Will Wang, Scott Counts, Siddharth Suri  
**arXiv:** [2507.07935](https://arxiv.org/abs/2507.07935)  
**Date:** 2025-07-10  
**Summary page:** https://www.summarizepaper.com/en/paper/2507.07935v1/

## Key points

- Authors Kiran Tomlinson, Sonia Jaffe, Will Wang, Scott Counts, and Siddharth Suri analyze the impact of generative AI on work activities and occupations.
- Understanding the effects of AI on the economy is a crucial societal concern due to rapid adoption and potential revolutionization of tasks across sectors.
- Common work activities where individuals seek assistance from AI include information gathering and writing.
- AI excels in providing information and assistance, writing content, teaching, and offering advice.
- The study introduces an "AI applicability score" for each occupation based on activity classifications, task success metrics, and impact scope.
- Knowledge-based professions like computer science and mathematics have high AI applicability scores.
- Office administrative roles and sales positions involving communication rank high in AI applicability.
- The research explores correlations between wages/education levels and AI applicability within occupations.
- Real-world usage data is compared with predictions regarding occupational impacts of AI technology.

## Summary

In their work titled "Working with AI: Measuring the Occupational Implications of Generative AI," authors Kiran Tomlinson, Sonia Jaffe, Will Wang, Scott Counts, and Siddharth Suri delve into the impact of generative AI on various work activities and occupations. With the rapid adoption of AI technology and its potential to revolutionize tasks across different sectors, understanding its effects on the economy has become a crucial societal concern. The study focuses on analyzing how people interact with AI in their work activities and which occupations are most involved in utilizing it. The researchers utilized a dataset comprising 200k anonymized conversations between users and Microsoft Bing Copilot, a generative AI system. Through this analysis, they discovered that common work activities where individuals seek assistance from AI include information gathering and writing. On the other hand, AI itself excels in providing information and assistance, writing content, teaching, and offering advice. By combining these activity classifications with metrics measuring task success and impact scope, the team calculated an "AI applicability score" for each occupation. Their findings revealed that knowledge-based professions such as computer science and mathematics scored highest in terms of AI applicability. Additionally, office administrative roles and sales positions involving communication and information dissemination also ranked high. The study also delves into successful types of work activities performed by AI systems. It examines correlations between wages/education levels and AI applicability within occupations as well as compares real-world usage data with predictions regarding occupational impacts of AI technology. Overall,this research sheds light on how generative AI is shaping various industries by enhancing productivity through automation while also highlighting potential implications for different job roles in today's evolving technological landscape.