Totally agree with this. The concerning part for me is that ML is about making a "best guess" given some data. This means that your tests may pass one time and

Totally agree with this. The concerning part for me is that ML is about making a "best guess" given some data. This means that your tests may pass one time and fail another - inconsistent tests aren't tests at all.