Entry

The Effects of Generative AI on High-Skilled Work: Evidence from Three Field Experiments with Software Developers

Zheyuan (Kevin) Cui, Mert Demirer, Sonia Jaffe, Leon Musolff, Sida Peng, Tobias Salz

Synopsis

Three RCTs (Microsoft, Accenture, Fortune-100, 4,867 devs combined) show 26.08% increase in completed tasks; gains larger for less-experienced developers.

Keywords

·field experiments ·Copilot ·productivity ·less-experienced

Open paper ↗ Report issue ↗

Related entries

Sea Change in Software Development: Economic and Productivity Analysis of the AI-Powered Developer Lifecycle

June 26, 2023 · arXiv
The Impact of AI on Developer Productivity: Evidence from GitHub Copilot

February 13, 2023 · arXiv
Impacts of Generative AI on Agile Teams' Productivity: A Multi-Case Longitudinal Study

February 14, 2026 · FORGE 2026
Evolving with AI: A Longitudinal Analysis of Developer Logs

January 15, 2026 · arXiv
Generative AI at Work

April 23, 2023 · Quarterly Journal of Economics 2025
Experimental evidence on the productivity effects of generative artificial intelligence

March 1, 2023 · Science 2023