I’m telling you. It’s fantastic for the boring and repetitive garbage. Databases? Oh hell yeah, it does really well on that, too. You have no idea how much I hate working with SQL. The ONLY thing it still struggles with so far is negative tests. For some reason, every single AI I’ve ever tried did good on positive tests, but just plain bad in the negative ones.
Yes, software, and specifically C# unit tests in my case.
Positive unit tests check if the code works as expected when given valid inputs. They confirm that the function or module behaves correctly under normal conditions.
Negative unit tests check how the code handles invalid or unexpected inputs. They ensure that errors are properly caught, exceptions are handled, and the system doesn’t break when things go wrong.
As for examples, it’s just the LLMs I have tried never wrote negative tests that actually worked. If you use Visual Studio, you’re probably familiar with those check marks that it has on unit tests. Those become green check marks when the test is valid, red X when it is invalid (isn’t correct). The negative tests from LLMs always have red X’s. Hope this makes sense.
I’m telling you. It’s fantastic for the boring and repetitive garbage. Databases? Oh hell yeah, it does really well on that, too. You have no idea how much I hate working with SQL. The ONLY thing it still struggles with so far is negative tests. For some reason, every single AI I’ve ever tried did good on positive tests, but just plain bad in the negative ones.
I assume we’re talking about software testing? I’d like to know more about:
The meaning of negative and positive tests in this context
Good examples of badly done negative tests by LLMs
Yes, software, and specifically C# unit tests in my case. Positive unit tests check if the code works as expected when given valid inputs. They confirm that the function or module behaves correctly under normal conditions. Negative unit tests check how the code handles invalid or unexpected inputs. They ensure that errors are properly caught, exceptions are handled, and the system doesn’t break when things go wrong.
As for examples, it’s just the LLMs I have tried never wrote negative tests that actually worked. If you use Visual Studio, you’re probably familiar with those check marks that it has on unit tests. Those become green check marks when the test is valid, red X when it is invalid (isn’t correct). The negative tests from LLMs always have red X’s. Hope this makes sense.