AI Coding Agents Inflate Code Volume Without Boosting Software Output

The promise of increased developer productivity via AI coding agents is hitting a critical operational wall: the human audit. While these tools accelerate the generation of code, they are not translating into a proportional increase in finished software, according to reporting from Ars Technica.

Researchers Fiona Chen and James Stratton from Harvard University analyzed data from Jellyfish, encompassing 300 million work events across more than 700 software development firms from 2021 through March 2026. Their findings indicate that the introduction of AI coding agents leads to a 30% increase in total lines of code, a 23% rise in pull requests, and a 20% increase in total commits.

However, this surge in volume has created a downstream crisis in the review process. Ars Technica reports that the average time between a pull request submission and its merge balloons by 49% after AI agents are introduced. Furthermore, the number of comments per pull request increases by 35%, and the share of pull requests requiring changes nearly doubles.

From an ROI perspective, these efficiency gains are effectively absorbed. The researchers found no statistically significant change in the resolution rate for Issues and Epics—the actual software features—tracked via tools like Jira. To manage the influx of AI-generated code, the share of workers performing code reviews increased by 14%, though the researchers noted no significant overall employment changes based on LinkedIn and Jellyfish data.

Despite 80% of firms using some form of AI code review by March 2026, the impact remains marginal. AI agents were responsible for only 10.8% of pull requests and 23.3% of review comments, leaving the vast majority of the auditing burden on human developers.