Component Networks for Explainable Artificial General Intelligence
We consider the mystery that a large part of scientific knowledge collected by mankind over centuries can be passed to a newborn child in about 20 years. We argue that the key is in the development and exploitation of components enabling linear proofs for the solutions of combinatorial problems. We put forth the notion of Crowdsourced Artificial General Intelligence (C-AGI) that exploits components of different backing databases. We use a putative assisted driving scenario and consider machine consistence seeking and self-training in a C-AGI system.