Slightly aside; I wonder ... How large is the difference between arriving at an answer by "using your gut", vs. arriving at the answer using a Machine Learning

Slightly aside; I wonder ... How large is the difference between arriving at an answer by "using your gut", vs. arriving at the answer using a Machine Learning system? Especially in cases where the ML system is a simulated neural net? In both cases you are using a neural net based optimizer, and in both cases it is hard to apply (post-hoc) introspection.