Should an automated HMI test be really detailed?

I'm working on some automated HMI tests for an application with python-selenium. I'm currently facing a dilemma concerning the level of detail my tests should have.

Maybe an example would be more relevant: in my application, I have some data tables you can scroll. You have several features, like sorting the data by a specific field, filtering, etc..

On the sorting feature, I want to be sure that what the user does causes the display of what I want, so I scroll the table, get the data and store them (Tab1), apply the sorting and scroll again the data displayed by the application (Tab2). Then, I sort the first table I stored (Tab1), but this time with calculations in my test, and compare it to the second table I stored (Tab2).

This is a pretty detailed test, and I'm not sure it really is the purpose of HMI testing, because it mixes HMI and Unit tests . Do you know if it's the right method to approach this kind of test, or I am doing wrong here ?