Has the ceiling of the software world been locked in place? AI-generated code doesn’t create anything new at all. It simply calls upon the technical legacy accumulated by generations of programmers—libraries, frameworks, toolchains. Without these foundations, AI wouldn’t be able to write even a single line of runnable code. But the incentive mechanism in the open-source world has completely collapsed: an engineer spends a lot of time contributing to an open-source library, yet the recognition and rewards they receive are disproportionate. Meanwhile, another project bloated with hundreds of thousands of lines of AI-generated code can become a superstar just through marketing. The lower limit of the software world can still be sustained for a while by the technical legacy left behind by the previous generation of programmers. But as the older generation of programmers gradually retires, how many of the younger programmers trained by AI coding will have the willingness and ability to maintain and contribute to these foundational libraries in the future? The next generation of programmers lacks sufficient training and motivation to take over, and decay will accelerate. The ceiling of the software world has, in fact, already been locked. Is this really the case? Can’t help but shiver.