One paradigm is worth a thousand rules: How can we teach computers compassion (and stop humans from using them to hurt and kill!) ? Artificial Intelligence, Psychology, and theories of mind.( from:http://www.epmg.kz/upload/content/File/MIT%20Press%20-%20Mind%20Design%20II%20%281997%29.pdf)
Many experiments (with Artificial Intelligence) have shown that categories appear to be coded in the mind neither by means of lists of each individual member of the category, nor by means of a list of formal criteria necessary and sufficient for category membership, but, rather, in terms of a prototype of a typical category member.
The most cognitively economical code for a category is, in fact, a concrete image of an average category member. (1977, p. 30)
One paradigm, it seems, is worth a thousand rules.
Reading the above text it appears the simplest way for artificial (or synthetic) intelligence to classify things is to stereotype them.