Cognitive AI
Summary
Cognitive AI represents a paradigm shift in artificial intelligence research, moving away from the “big data for small tasks” approach towards a “small data for big tasks” model. This new paradigm aims to develop AI systems with humanlike common sense, capable of solving a wide range of tasks with minimal training data. The core of this approach focuses on five domains: functionality, physics, intent, causality, and utility (FPICU), which are collectively referred to as the “dark matter of vision.” These domains go beyond the traditional “what and where” framework of visual understanding, addressing the “why and how” aspects that drive the creation and development of visual scenes. By incorporating FPICU, cognitive AI systems can tackle complex challenges such as tool use, planning, utility inference, and social learning, demonstrating a more robust and adaptable form of artificial intelligence that more closely mimics human cognitive abilities.