The use of AI in the analysis of social relationships shows particularly impressive results. Two ways of working can be distinguished: the path of individualization and that of generalization. While one emphasizes what is special – what separates – , the path of generalization focuses on what is common. Although the two are not necessarily in conflict with each other, the generalization route has proven to be more efficient. There are reasons for this:
85% of individual behavior can be explained collectively, and a few dozen overlapping factors will suffice
The 15% of cultural and unique influencers breaks down into a multitude of individual aspects that are highly situational. They are more logistical than content-related (who gets what information and when?)
In other words: when working with AI, we aim to classify individual cases as efficiently as possible. We look at groups with typical characteristics and can assign them to individual cases. We can therefore say for the individual case what general properties it has, and we abstract from the individual ones. This is not so much a question of data security or morality, but of the economics of data flow and process organization. Global beats local optimum, as AI teaches us.
