You need some understanding of maths to realise the result you got was completely wrong cause you made a mistake pressing the buttons. Like the people who blindly trusted GPS instead of paying any attention and ended up with vehicles upside down in strange places.
A lot that goes into automation. When I did my PhD, I had equipment with a bewildering number of switches and dials which set the parameters of the data recording, and you'd better know what they all do otherwise your experiments would be a pile of crap. Step forward 10 years, a lot of the adjustments could be done by software by using an auto function. This is all well and good, but it's not perfect - if something was wrong, in order to try to make things work, the software could apply some crazy parameters. As a postdoc, I realised from talking to some later PhD students they had very little idea what all these parameters meant, because they just hit the auto function. Chances are some of their data was seriously screwed.