It is usually interpretable.

None of the above.

It uses reasoning to produce new knowledge from pre-defined knowledge.

It is usually impossible to explain how exactly it works.

It uses explicit representation of knowledge.