AI isn't making developers more productive – it's making them busier
5 hours ago
- #Code Review
- #Software Development
- #AI Productivity
- AI-coding tools have significantly increased code generation, with a 741% rise in lines written, but this has not translated into proportional productivity gains in software releases.
- Task-level productivity boosts from AI tools diminish sharply as work moves up the hierarchy, with only a 20% increase in releases, highlighting bottlenecks in code review, integration, and release processes.
- The role of developers is shifting from writing code to evaluating it, with review now consuming about half of their time, requiring organizations to adapt processes to manage increased AI-generated output.
- Despite a surge in new application releases due to AI assistance, total app usage remains flat, indicating many fail to attract audiences and underscoring that more software doesn't inherently create value.
- Engineering managers must rethink workflows to address constraints in the development pipeline, focusing on evaluation and testing as primary constraints to leverage AI tools effectively without overwhelming teams.