AI isn't making developers more productive – it's making them busier (leaddev.com)

Recent research from the Massachusetts Institute of Technology and the Wharton School highlights a critical paradox in the use of AI in software development: while AI tools have significantly increased the volume of code written—up by 741%—the actual productivity gains in terms of software releases are minimal, at just 20%. The bottleneck lies not in writing code but in the subsequent processes like code review and integration, which remain heavily reliant on human effort. As a result, despite the apparent acceleration in coding capabilities through tools such as Fable and Claude Code, developers find themselves busier, grappling with a growing pile of unfinished work. This shift necessitates a reevaluation of the developer's role—from code writer to code evaluator. Engineering managers are urged to adapt their processes, focusing on enhancing review and testing methodologies to keep pace with the influx of AI-generated code. Without these changes, the benefits of increased coding efficiency may be undermined, leading to a stagnant software release landscape and negligible user engagement with new applications. The findings underscore the complexity of integrating AI in development workflows and challenge the tech community to innovate not just in code generation, but in maximizing overall productivity through process optimization.