Learning is the ability to solve again — DedSec Project

Knowing how to locate a solution is real problem-solving. It is not identical to remembering a language's syntax or independently proving every algorithm correct. Use AI, documentation, and experiments without confusing the roles.

A good cycle:

1. Explain the problem in your own words.
2. Predict a result before running anything.
3. Try a small version.
4. Read errors instead of replacing the whole script.
5. Ask an assistant for why, alternatives, and failure cases.
6. Change the example yourself and predict the new result.
7. Rebuild without relying on the exact previous answer.

Example: a Python list unexpectedly keeps values between calls. Investigate mutable default arguments, reproduce with two calls, replace the default with None and allocate the list inside. Then explain why the function object preserved state.

Understanding test: can you adapt the fix to another function? Can you describe when it would not be needed?

AI is a useful collaborator, but generated code still needs review, tests, licensing checks, and an owner who understands its effect.