This report describes the principal findings and recommendations of a 2-year Rand research project on machine-aided knowledge acquisition and discusses the transfer of expertise from humans to machines, as well as the functions of planning, debugging, knowledge refinement, and autonomous machine learning. The relative advantages of humans and machines in the building of intelligent systems are explained. Background and guidance is provided for policymakers concerned with the research and development of machine-based learning systems. The research method adopted emphasized iterative refinement of knowledge in response to actual experience; i.e., a machine's knowledge was acquired initially from a human who provided enough concepts, constraints, and problem-solving heuristics to define some minimal level of performance. Sixty-two references are listed. (Author/FM)