Code Review Has to Change for AI-Generated Code
AI coding tools change the bottleneck in software engineering. For years, the scarce resource was implementation time. A team could only produce as much code as its engineers could write, debug, and locally validate. AI coding assistants weaken that constraint. They can produce features, tests, migrations, refactors, and glue code much faster than a human team can type them. That sounds like a productivity win, and often it is. But the downstream system still has to absorb the code. Somebody still has to understand the change, verify the behavior, check security boundaries, reason about failure modes, integrate it with the rest of the system, deploy it safely, and operate it after merge. ...