Brought to you by the same AI that fixes tests by removing them.

If a test fails but is never called, did it ever fail at all?

If you use "AI" to lump together all the models, then sure.