>>10996793no it's really not.
First of all, solving these sorts of sums is how the integral is derived in the first place.
Second of all, what you are describing will be a series of boxes of width 1 which are lined up with the curve.
So you are approximately right, but not right.
Your idea is used in statistics quite often when you are approximating.
this looks very close to but it is only an approximation.
The error is on the order of ln(n)