Github Copilot Study

GitHub’s Copilot AI Coding Assistant Boosts Developer Productivity and Happiness: Report

Software developers using GitHub’s Copilot AI assistant are likely to code faster and feel “more fulfilled with their job” than those who don’t, according to GitHub’s new report. GitHub timed the survey of more than 2,000 programmers to the anniversary of the Copilot technical preview though fewer than three months after its public debut.

Copilot Pal

GitHub Copilot is a “pair programmer,” with the AI taking a junior role in a partnership that traditionally has two human developers coding the same project simultaneously. Human pair programmers collaborate with their partners by offering ideas, spotting errors, and generally improving the final project. Copilot, descended from the specialized Codex form of OpenAI’s GPT-3 language model, performs the same role, but as a digital presence only. Copilot can adapt to the programmer’s style in order to suggest lines of code, solutions to problems, and possibly useful features. According to the survey, Copilot is more than living up to its goals.

On the more qualitative front, a majority of the Copilot users said they were “less frustrated when coding” and “more fulfilled with my job” at 59% and 60%, respectively. And 74% said Copilot enabled them to “focus on more satisfying work,” presumably because the AI helped dispatch the more tedious and repetitive elements of coding. Nearly every programmer said that Copilot helped them complete tasks more quickly, and an experiment comparing programmers completing a task with or without using Copilot bore that feeling out. Completing a programming task took more than twice as long, 161 minutes on average, for the developers coding without Copilot, compared to the 71-minute average of the group using Copilot. Not bad for a $10 a month or $100 a year subscription. Plus there’s a 60-day free trial, and it’s free to verified teachers, students and approved open-source software managers.

“The takeaway from our qualitative investigation was that letting GitHub Copilot shoulder the boring and repetitive work of development reduced cognitive load. This makes room for developers to enjoy the more meaningful work that requires complex, critical thinking and problem solving, leading to greater happiness and satisfaction,” GitHub researcher Eirini Kalliamvakou explained.”There’s more to uncover! We’re conducting more experiments and a more thorough analysis of the experiment data we already collected—looking into heterogeneous effects, or potential effects on the quality of code—and we are planning further academic publications to share our findings.”


GitHub’s AI Coding Assistant Copilot Launches

Amazon’s New CodeWhisperer AI Programming Assistant Challenges GitHub Copilot

Microsoft Brings GPT-3 to Natural Language Programming Tool