The article discusses the challenges and frustrations of working with Large Language Models (LLMs) in software development, particularly the feeling of burnout and the loss of creative control. The author describes the experience of using LLMs as a double-edged sword, where the benefits of increased productivity are offset by the fatigue of supervising and reviewing the output. The article also touches on the idea that the skills required for software development are shifting, with a greater emphasis on high-level thinking and judgment. The author concludes that the humans are still in the loop, but they are tired and need to adapt to the new reality.