LLMs and performative productivity

This article critically examines the author's experience using AI agents for software development tasks, questioning whether the apparent productivity gains are genuine or merely performative. The author describes completing tasks like codebase contributions, upgrades, and bug fixes faster than ever, but then realizes they lacked deep understanding of the code, context, and real user impact. They note abandoned projects, unnecessary tests, and a net loss of personal skill development. The piece explores the trade-offs between speed and meaningful learning, ultimately asking if AI-driven productivity is a net gain or just busywork.