Representation as a fluent: An AI challenge for the next half century

The characteristics of autonomous agents that can solve multiple and evolving goals in a complex, messy environments in collaboration with some agents and in conflicts with others, are discussed. To achieve the ambitious applications, the representations must be a fluent, and must evolve under machine control. Automatic representation development, evolution, and repair must be a major goal of AI research for the years to come. A reasoning system must able to develop, evolve, and repair their underlying representations as well as reasons with them. If a reasoning system is restricted to a fixed ontology, it will rapidly get out of date. McNeill's approach assumes that there has been an attempt to standardize ontologies, and with any sufficient large agent community, small differences between the ontologies will be inevitable. McNeill's ontology refinement system (ORS) identifies and repairs these ontological mismatch at runtime.