Embodied/Situated Cognition and Evolutionary Devolopment as They Apply to Artificial Intelligence

Much of the work in artificial intelligence (AI), automated planning, machine learning, computer vision, and related fields focus primarily on the capabilities of formal logic, statistics, and other mathematical tools to emulate human cognitive capabilities. Though these areas have in fact seen significant success (e.g., spam detection, information retrieval, etc.), the much-sought goal of a human-level, strong AI has been firmly outside the reach of the field. In the past two decades, however, seekers of strong AI have gradually been waking up to a new paradigm: it may not be possible to replicate human-level cognitive capabilities divorced from the human physical experience of the real world. Restated, human-level reasoning cannot be achieved using the mechanics of formal logic, probability, and statistical inference alone; human-level cognition instead requires an approximation of human physical and social experience in which to learn.