I’ve published a new pair of articles on Ask Good Questions.
This week’s pair came out of something I’ve been seeing repeatedly while working with AI on larger software projects.
Sometimes the AI makes the wrong fix. But when you look closely at why it made that choice, the mistake can tell you something important about the architecture itself.
If a capable AI worker can enter a project, follow what looks like a reasonable path through the code, and end up changing the wrong layer or the wrong implementation, it’s worth asking whether a new human programmer might make the same mistake.
That doesn’t excuse the bad fix.
But it may expose places where ownership isn’t obvious, where old and new implementations coexist, or where too much of the architecture depends on history and tribal knowledge instead of being clear from the code itself.
The Guide, AI Didn’t Create the Architectural Problem. It Exposed It., looks at that broader idea.
The matching Field Note, The Wrong Fix Was the Right Warning, tells the story behind it, including how a wrong local fix in one of our projects turned into a much more useful warning about the way responsibilities were divided across the system.
Guide: AI Didn’t Create the Architectural Problem. It Exposed It.
Field Note: The Wrong Fix Was the Right Warning
Charles