AI coding agents increase code review time, researchers find

AI coding agents, while increasing the volume of code generated, have inadvertently lengthened the software development lifecycle due to a significant rise in code review time and effort. Researchers found that the average time between submitting a pull request and merging it into the codebase increased by 49% after the introduction of these AI tools. This extended review process is evidenced by a near doubling in the share of pull requests that require changes and a 35% increase in the number of comments per pull request.

In response to the increased workload, the study observed a 14% rise in the proportion of workers dedicated to code reviews. Despite the widespread adoption of AI for code review, with 80% of measured firms utilizing it by March 2026, human oversight remains dominant. AI agents accounted for only 23.3% of review comments and 10.8% of all pull requests, indicating that humans still handle the bulk of this critical stage.

While AI coding agents are still evolving, and firms are learning to optimize their deployment, the current trade-off between coding speed and review time presents a challenge. The researchers noted that significant employment changes were not directly attributable to AI adoption based on their data. The study, which included data up to March 2026, suggests that the benefits of AI coding agents may not yet outweigh the increased human effort required for code verification for many companies.