Solving a Problem With or Without a Program

To solve a problem, an ordinary computer system executes an existing program. When no such program is available, an AGI system may still be able to solve a concrete problem instance. This paper introduces a new approach to do so in a reasoning system that adapts to its environment and works with insuffcient knowledge and resources. The related approaches are compared, and several conceptual issues are analyzed. It is concluded that an AGI system can solve a problem with or without a problem-specific program, and therefore can have human-like creativity and exibility.