AI use cases › Technology

Microsoft, Accenture and a Fortune 100 firm: Three randomized field trials of Copilot with 4,867 developers

A large study of 4,867 developers at three big companies, testing whether an AI coding helper improves output.

How it works

Some developers were randomly given the AI helper and others were not, and researchers compared the work finished each week.

What it delivers

Users completed about 26% more tasks weekly, with bigger gains for junior developers; the results are not yet peer reviewed.

Modelled business case

Figures are a model built from published benchmarks for this type of AI, not the organisation's own numbers. Change any assumption in the app.

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

Create your free Intlectify account to bookmark use cases, model the business case on your own numbers and start the free AI course.