AI isn’t making developers more productive – it’s making them busier
SMRTR summary
A 741% surge in code written. Just a 20% increase in software actually shipped. That gap tells you everything about the AI coding revolution right now.
A study of more than 100,000 GitHub developers, conducted by researchers at MIT and the Wharton School of Business, finds that AI tools are making developers busier, not necessarily more productive. The bottleneck isn't writing code anymore. It's everything that comes after.
Code review, integration, testing, releases - those still depend on human judgment, and human capacity hasn't scaled the way AI output has. The result is a traffic jam deep in the pipeline.
"A system's throughput is set by its slowest stage," says Alfonso Graziano of software developer Nearform. Push code in faster, and work simply piles up at the constraint.
The implications run all the way to consumers. App releases have spiked, but total app usage remains flat, suggesting much of this AI-assisted software is finding no audience at all.
SMRTR provides this summary for quick context. The original article belongs to Reddit.
Read the original article