# Common-Sense Reasoning **Domain:** Artificial Intelligence / Knowledge / Physical World **Doc Type:** Concept Node **Maturity:** Developed **Related:** [[Marvin Minsky]], [[Leela AI]], [[Inference]], [[World Modeling]], [[Artificial Intelligence]] --- ## Definition **Common-sense reasoning** is the capacity to draw useful conclusions from the ordinary structure of the world: objects, bodies, causes, intentions, time, space, social expectations and the unstated background that human beings usually take for granted. In artificial intelligence, the problem is difficult because much of this knowledge is contextual, defeasible and acquired through embodied interaction rather than stated as universal rules. ## Leela AI Context [[Leela AI]] approaches the problem through visual observation, infant-learning schemas and a hybrid of pattern recognition and [[Causal Learning|causal learning]]. Its manufacturing environment supplies recurring physical actions from which a system can build grounded relationships. ## Key Insight **Common sense is not a small set of obvious facts; it is the vast background that makes explicit reasoning possible.**