11–16 of 16 posts
Anyone made the opposite product? Human written tests, that an AI tries to make green?
I'd expect that the ai written code doesn't do the right thing outside of the given test cases
If I imagine a workflow of
1. Updating some code (with a regression) 2. Running the ai to generate tests 3. The tests pass
How do I know that the ai didn't assume the regression is expected? Do I read the explanation of the test cases?
I don't really understand what the test cases will do? Is the code expected to already correct, and the tests are documentation? If I imagine a workflow of 1. Updating some code (with a regression) 2. Running the ai to generate tests 3. The tests pass How do I know that the ai didn't assume the regression is expected? Do I read the explanation of the test cases?
They are generated by AI, so unit tests will be implemented by case definitions.
Anyone made the opposite product? Human written tests, that an AI tries to make green?
IMHO, Testing is one thing that needs to be inspected when it comes to AI code generation
You have full control over generated tests. Also, I put focus on putting as much control as possible, so you can set different inputs to reach higher quality of generated tests.
At the end, you add or append file with generated tests, so you have full control over them.