What needs to be added to machine learning?

The question we ask is how to build on the success of machine learning to address the broader goals of artificial intelligence. We regard reasoning as the major component of cognition, other than learning, that needs to be incorporated. We suggest that the central challenge therefore is to unify the formulation of these two phenomena, learning and reasoning, whose conventional formulations are contradictory, into a single framework with a common semantics. We propose Robust Logic for this role, as a framework with a satisfactory theoretical basis. Testing it experimentally on a significant scale remains a major challenge for the future.