An inference engine makes a decision from the facts and rules contained in the knowledge base of an expert system or the algorithm derived from a deep learning AI system. The inference engine is the processing component in contrast to the fact gathering or learning side of the system.
In years past, the inference engine referred to software-only expert systems. Today, there are specialized chips for executing in parallel the enormous number of calculations in a neural network, which is the primary AI architecture. See neural network
, deep learning
and expert system