There are two main perspectives in the world of automated redistricting that differ based on their use case. The first perspective is using automated algorithms to replace the work of human map drawers by creating an optimized map given certain criteria. The most common type of optimization algorithms maintain contiguity and population balance of the districts and try to maximize the “compactness” through some measure of shape.

The second perspective is using automated algorithms to perform analytical work by creating large numbers of redistricting plans which are called ensembles. Someone with this perspective, would write an algorithm to generate thousands of maps to analyze something like the effects of gerrymandering on past elections or where one particular map stands relative to other possible maps. 

Even though ensembles draw multiple maps and one map could be selected to replace the work of a human map drawer (and similarly one could draw multiple optimized maps), these two perspectives are often separated because of the different algorithms used to draw their maps.