Because if I learned one thing it‘s that eye balling a solution doesn‘t work. In „classical“ code you can test edge cases and you‘ll know „it works“, but with ML you really need good coverage and out of sample testing.