AI usage generates a fragile codebase

What happened: A developer laid off from Meta in May describes seeing brittle, AI-generated code from peers, and now at another large tech company generates 100% of code with AI as expected. Why it matters: The author argues AI-assisted development produces fragile codebases, citing shallow tests, overly complex code, declining human understanding, and faster pull requests that erode review quality. Key numbers: Author says he generated at least 90% of his code using AI in his last months at Meta; AI did not 10X or even 2X his output.

A developer laid off from Meta in May describes seeing brittle, AI-generated code from peers, and now at another large tech company generates 100% of code with AI as expected.

The author argues AI-assisted development produces fragile codebases, citing shallow tests, overly complex code, declining human understanding, and faster pull requests that erode review quality. Key numbers: Author says he generated at least 90% of his code using AI in his last months at Meta; AI did not 10X or even 2X his output.