AI agents are great at writing new code — but then the human bottleneck slows everything down
While AI has boosted the amount of code we can now produce, it needs substantially more reviewing.
Recent research from Harvard University has revealed that AI agents significantly boost coding productivity, with developers able to produce up to 30% more code. However, this efficiency is counteracted by the human bottleneck of code review. Code commits and pull requests have increased by 20% and 23% respectively, while comments per pull request have surged by 49%.
This has led to a significant rise in the time taken for code review, from an average of 7.03 days to 10.48 days. The researchers, Fiona Chen and James Stratton, found no statistically significant improvement in the quality of completed software despite the increased output. The bottleneck is now code review, with a 49% uptick in review time, 35% more reviewer comments per pull request, and a 14% increase in employees conducting reviews.
While AI can generate code more quickly, it requires substantial corrections, leading to the question of whether AI is truly beneficial. Despite the shift in bottlenecks, AI adoption has not resulted in significant job losses, suggesting that the real issue lies in how companies can effectively utilize AI to support employees and minimize unnecessary workload, rather than replacing workers entirely.
Written by urgent.news from TechRadar's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
