Computational Awareness: Another Way towards Intelligence

摘要

Artificial intelligence (AI) has been a dream of researchers for decades. In 1982, Japan launched the 5~(th) generation computer project, expecting to create AI in computers, but failed. Noting that logic approach alone is not enough, soft computing (e.g. neuro-computing, fuzzy logic and evolutionary computation) has attracted great attention since 1990s. After another 2 decades, however, we have not got any system that is as intelligent as a human, in the sense of "over-all performance". Instead of trying to create intelligence directly, we may try to create "awareness" first, and obtain intelligence "step-by-step". Briefly speaking, awareness is a mechanism for detecting any event which may or may not lead to complete understanding. Depending on the complexity of the events to detect, aware systems can be divided into many levels. Although low level aware systems may not be clever enough to provide understandable knowledge about an observation; they may provide important information for high level aware systems to make understandable decisions. In this paper we do not intend to provide a survey of existing results related to awareness computing. Rather, we will study this field from a new perspective, try to clarify some related terminologies, and propose some problems to solve for creating intelligence through computational awareness.