Section 063 · Chapter 9, Generated Code Changes the Job

AI-Generated Code That Looks Right but Is Wrong

AI-generated code often fails in a dangerous way: it looks clean, compiles, and still implements the wrong behavior.

generated codegenerated code looks right wrong

What to do

  1. Do not only test the happy path the prompt described.
  2. Test the neighboring cases the prompt did not mention.
  3. Use property-based tests, metamorphic tests, boundary matrices, and requirement-to-test traceability.

Evidence to preserve

  • Preserve the inputs, versions, configurations, raw outcomes, and results for generated code, generated code looks right wrong needed to reproduce work on AI-Generated Code That Looks Right but Is Wrong.
  • Report results for generated code, generated code looks right wrong by relevant slice, separate blocker failures from averages, state uncertainty and blind spots, and connect the result to a release decision.

Expert note

Use property-based tests, metamorphic tests, boundary matrices, and requirement-to-test traceability. AI-generated code should be judged by behavioral evidence, not by whether it looks idiomatic.

Continue the conversation

Apply this to your context.

Save your product context once, then open a focused conversation that combines it with this concept.

Cite this page

Jason Arbon. "AI-Generated Code That Looks Right but Is Wrong." Testing AI Knowledge Edition, section 63.

https://jarbon.ai/testing-ai/knowledge/ch063-generated-code-looks-right-wrong.html

Shared across the Knowledge Edition

Adapt every concept to your world.

Describe your product, role, users, risks, constraints, or current quality problem. This stays in this browser until you choose to send it to ChatGPT.

Saved only in this browser.0 / 2400