The purpose of writing tests is to uncover as many bugs as possible by actively seeking to break the code. Merely verifying that the software performs expected tasks, is insufficient. Yes you should test and confirm that the software operates as intended in a perfect utopian scenario, but only thinking about tests this way is highly myopic, it's only a matter of time before it breaks at the hands of users, instead of your development team.
A failing tests is not a test that errors out and runs red. A failing test is a test that failed at doing it's job which is detecting bugs, it ran green this time, it didn't find any errors, it failed.