Who Made This Decision?
We built a product that, at the old pace, would have taken a year before there was anything to show. We had it in two months. It worked, it looked good, and every feature that was asked for showed up. Then it was time to evaluate a small change. It meant touching 80 files. Not because the change was big, but because the code had no structure. It was one big mass that had grown with a single…
The article "Who Made This Decision?" discusses the impact of AI-generated code on software development and the lack of oversight in the decision-making process. The author argues that while AI can write code quickly and efficiently, the absence of human review leads to a multitude of problems.
Key points include:
1. The product described took only two months to build, despite being a complex feature requiring 80 files of code changes. This rapid development is attributed to AI's speed and accuracy in generating code, but the process often overlooks important considerations.
2. The codebase exhibits poor structure, with dead code, scattered stores, and a lack of abstraction separating functionality from implementation details. This hampers maintainability and makes future changes more difficult.
3. Thousands of tests were written, many of which check for simple conditions like the presence of a CSS class. These tests become brittle when the UI changes and essentially act as a barrier to progress.
4. The author attributes the poor quality of the code to the lack of oversight in the decision-making process, rather than the AI itself. AI can generate code quickly, but it doesn't have the judgment or foresight to make informed decisions about code structure and design.
5. The article emphasizes the need for human involvement in deciding architectural decisions and abstractions, as AI tends to prioritize quick delivery over long-term maintainability.
Overall, the piece argues that the real issue is not who wrote the code, but who made the decisions about how that code should be structured and designed. Without proper oversight and human judgment, AI-generated code can lead to a tangled, hard-to-maintain codebase.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
