Generative artificial intelligence has revolutionized how we approach various tasks, especially in programming, where “vibecoding” (AI-generated code) has become a common practice. While some view it as a radical improvement in productivity, its limitations become starkly apparent in complex projects. A clear example is “vib-OS,” an operating system developed with AI code that demonstrates failures even in the most fundamental functions.
AI can produce in seconds what previously took weeks or months, from digital artwork to books. This capability has led to increasing dependency, with people consulting AI before searching for information themselves online, completely transforming the work paradigm, including in programming. This has impacted sites like Stack Overflow, which are experiencing a decline as users prefer the immediacy of AI.
Despite its imperfections and propensity for errors, reliance on AI continues to grow. YouTuber “tirimid” decided to test the limits of AI by attempting to create a complete operating system. The result, vib-OS, was a disaster from the outset. The system encountered problems booting due to incorrectly functioning AI-generated scripts. After more than an hour of attempts, it finally managed to start.
vib-OS was subjected to nine fundamental tests, of which it only passed five. It utterly failed internet connectivity, unable to connect to any network or even diagnose network issues. Its “macOS-inspired graphical interface,” while simple, also showed deficiencies: the file explorer worked only in a limited capacity, and the text editor couldn’t save documents. Furthermore, it lacked Python support and couldn’t execute external programs.
Most surprisingly, it was unable to run DOOM, the iconic 1993 game known for its almost universal compatibility (it runs even on a tractor with Linux). The only game that managed to function was Snake. The conclusion is clear: vib-OS, though it boots and has an interface, is incomplete, and the AI-generated code is riddled with errors. For the curious, the project can be found on viralcode’s GitHub.
