Renews.
← AI

Harvard study: AI coding agents boost output but slow review

A Harvard working paper finds that while AI coding agents increase code volume, they significantly lengthen review times without raising finished work.

AI·1 source
Photo: Pexels / Pixabay

Harvard Ph.D. candidates Fiona Chen and James Stratton published a working paper analyzing the impact of AI coding agents on engineering firms. The study, dated August 4, 2026, uses data from January 2021 to March 2026. Engineers at adopting firms added approximately 30 percent more code, with lines of code rising by 1,494 per worker-month. Commits increased by 20 percent and pull requests by 23 percent. However, the amount of finished work did not measurably rise, with resolved issues moving by only 0.12 per worker-month.

Getting code through review took 49 percent longer after adoption. The time from pull request submission to merge rose by 3.45 days. The share of pull requests receiving a formal change request increased by about 12 percentage points, and review comments per pull request rose by 35 percent. The share of workers doing any code review rose by 14 percent. By March 2026, nearly 80 percent of firms had used AI review tools. Despite this, AI generated only 23.3 percent of review comments, and only 10.8 percent of pull requests received even one AI comment.

The study estimates rule out a decline in overall employment larger than 2.9 percent. A noisier estimate rules out a decline in engineering employment larger than 13.8 percent. The data source includes the Jellyfish analytics platform and LinkedIn employment records from Revelio Labs.

Sources

Every sentence of this story was checked against facts quoted from these articles. How we make the news

Spotted an error? Tell us

Sign in

Keep the stories you like in your favorites.

Forgot your password?

No account?