Perspectives on artificial intelligence: Three ways to be smart

Abstract

Three different styles of achieving Artificial Intelligence (AI) are discussed and compared. The earliest, and best-known to the general public, is the computational approach to AI that takes the brain to be some form of computer. This focuses on a narrow form of intelligence, abstract reasoning. By contrast, Artificial Neural Networks sees the brain as a brain rather than computer. This focuses on categorization and pattern recognition. The newest perspective is Evolutionary Robotics. This takes the broadest view of intelligence and cognition, seeing it as adaptive behaviour in a physical world; cognition and the mind are not centred in the brain at all. Rather than three different methods of achieving similar goals, these perspectives are aimed in very different directions. The implications of this for developing artificially intelligent SmartData agents are discussed.