Harvard study finds AI coding tools add code without raising software output
Overview
Harvard researchers Fiona Chen and James Stratton find little evidence that AI coding tools increase software output or reduce employment at firms that use them.
Their study uses aggregated analytics from Jellyfish, which tracks engineering team output, covering more than 700 software firms.
The authors say efficiency gained during the coding stage is absorbed by later steps. As code review takes longer, pull requests need more revisions and reviewers leave more comments, which the study identifies as the main bottleneck.
Written by AI from the articles below · updated Oct 9, 5:00 PM ET
Check the sources:
Article timeline
The articles in this story. Times are ET.
- Ars Technica · AINewsAI coding agents generate more code but not more software, study finds
AIA study by Harvard researchers Fiona Chen and James Stratton, using Jellyfish engineering data from over 700 software firms, finds little evidence that AI coding tools increase software output or reduce employment. The authors report that efficiency gained during coding is absorbed by downstream constraints, mainly longer code review, more pull request revisions, and more reviewer comments.
Heat trend
Not enough continuous observations to show a trend yet.