Updated

A new study found that while AI coding agents are generating significantly more code, this hasn't translated into proportionally more finished software, because efficiency gains are being absorbed by the human review process, which remains a bottleneck. The research challenges assumptions that AI coding tools straightforwardly accelerate software development. The findings add nuance to the debate over AI's impact on developer productivity, suggesting that organizational and review processes—not just code generation speed—determine how much value AI coding assistants actually deliver.