reception to follow
AI has never quite come to grips with how human beings reason in daily life. Many deep learning and automated theorem-proving technologies exist, but they cannot serve as a foundation for building interactive systems that reason automatically. In this talk, I will trace their limitations back to two historical developments in AI: the motivation to establish automated theorem-provers for systems of mathematical logic, and the formulation of nonmonotonic systems of reasoning. I'll then describe properties of human reasoning that cannot be simulated by current machine reasoning or machine learning methodologies. People can generate inferences on their own instead of just evaluating them. They use strategies and fallible shortcuts when they reason. The discovery of an inconsistency does not result in an explosion of inferences -- instead, it often prompts reasoners to abandon a premise. And the connectives they use in natural language have different meanings than those in classical logic. Only recently have cognitive scientists begun to implement automated reasoning systems that reflect these human patterns of reasoning. A key constraint of these recent implementations is that they compute, not proofs, probabilities, or truth values, but possibilities.