Cost of 'saying yes' in AI era: coding cheaper, owning code not

A GitHub blog post argues that while AI reduces the cost of writing code, the cost of owning code—including maintenance, understanding, debugging, and security review—has not decreased. It proposes a framework to help teams gauge which code changes are truly cheaper in the AI era. For example, AI-generated code that is frequently modified or in critical paths may have higher long-term ownership costs than hand-written code. For developers using AI coding tools, this means not just celebrating faster generation but reconsidering tech debt and code quality. Teams need extra review and testing for AI-produced code, or risk accumulating an unmaintainable codebase. The article suggests classifying changes as 'cheap' or 'expensive' and optimizing workflows accordingly.