AI Got Better While I Was Away — But Did We Get Dumber? A Developer's Honest Audit of What AI Actually Replaced vs. What It Just Made Lazy

Description and use

AI boosts productivity for boilerplate and documentation, yet can erode developer skills if used as the primary author. The "write first, ask second" practice preserves understanding and improves quality.

Problem

Developers risk losing foundational understanding when AI assistants replace manual coding and debugging, leading to lazy patterns that erode skill.

Outcome

AI improves productivity for boilerplate and routine tasks but can erode developer understanding if used as a primary author. A productive pattern—write first, ask second—maintains skill while leveraging AI.

Workflow

• 1. Write initial code for non-trivial logic or architecture.
• 2. Ask AI to review, suggest improvements, and identify edge cases.
• 3. Question AI explanations before applying fixes.
• 4. Audit AI-generated code for type safety, casts, and assumptions.
• 5. For junior developers, enforce "write first" for first three months and require explanation of pasted code before shipping.

Evidence

Two years ago, I could not make a Copilot suggestion pass code review without rewrites. Today, half my PRs ship with AI-generated code that I barely edited.

Source

Debugging has become search-and-paste. When an error occurs, the reflex is no longer "let me trace through the call stack and understand why this happened." It is "let me ask the AI what this error means and what to change."

Source

Type safety is being bypassed. When you paste AI-generated code, you skip that engagement. The code compiles. The types are wrong.

Source

I write first, then ask. For anything non-trivial — any function with business logic, any architectural decision, any code that touches shared state — I write the initial version myself. Then I ask the AI to review it.

Source