If i had to write the tests myself I would never try to get 100% test coverage or anywhere close, I would just test the "important" parts

With AI I can still manually write the important tests, but I can also get it to generate the mundane tests for making sure methods take different types of arguments and still return the expected result