Evan X. Merz · 2026-11-07 · notable
Evan X. Merz — AI usage generates a fragile codebase
Evan X. Merz, who wrote at least 90% of his code with AI at Meta, argues that AI-built SaaS codebases get fragile: shallow tests, over-complex code, weaker human understanding and more pull requests than reviewers can check.
A former Meta engineer explains why code written mostly by AI breaks more often.
What is it?
Evan X. Merz's essay 'AI usage generates a fragile codebase' draws on his time at Meta, where generating at least 90% of his code with AI was the minimum for a decent rating. He says AI did not even double how much code he wrote.
How does it work?
Merz names four causes. AI writes shallow tests, such as trivial getter and setter checks; it adds needless helper methods and over-applies rules like DRY; developers who lean on it understand their code less; and AI pull requests arrive faster than reviewers can check them properly.
Why does it matter?
The result, in Merz's view, is more incidents, slower recovery because nobody knows the code well, and bugs that only real users find. He argues that AI solves problems that fit inside its context window but ignores long-term ones, so products for people still need human judgment.
Who is it for?
engineering managers and teams with AI coding mandates