The Real Value of GitHub Copilot Rubber Duck

This article discusses GitHub Copilot Rubber Duck, a new feature in GitHub Copilot CLI that introduces a second, different AI model to review code plans and implementations. It highlights how AI-assisted coding often produces polished but wrong outputs, and that self-review is insufficient. By having a separate model challenge the original work, it adds friction to catch blind spots, especially in complex, multi-file tasks. The article argues this is a healthier direction than simply generating more code, emphasizing that more AI doesn't mean better engineering. It explores why this matters at platform scale and how it addresses subtle failures in AI-generated code.