Simon Willison's demonstration, while neat, arguably highlights a core tension in AI-assisted development. The generated component solves a narrow, pre-defined problem but may lack the robustness, error handling, and extensibility of a hand-crafted solution. The real test for such tools will be their ability to handle edge cases and integrate into complex build systems, areas where current AI code generation often appears to fall short.
This experiment underscores that the immediate value may lie in rapid prototyping and idea generation, not in production-ready code, a distinction that will likely determine the tool's practical adoption.