When a training program fails to markedly influence the development of high-tech complex skills (such as electronic troubleshooting), the failure can generally be traced to two sources. First, failure occurs when training is not based on clear and explicit models of the desired expertise. For problem solving expertise, specifications of the expert's internal strategic processes for handling complex problems and the particular forms of knowledge and skill that support the strategies are especially critical. Secondly, failure occurs because the training of complex mental skills often fails to consider the conditions that are needed for the development of cognitive expertise, though similar conditions for the development of advanced physical skills are well known. They include extensive, constructive practice sessions where "the game is played" (i.e., authentic problems are solved) under realistic conditions. For such practice to be constructive, the trainee needs commentary and guidance from a coach who, among other things, can model the desired (problem solving) performance and carefully sequence problems according to the trainee's progress, while at the same time providing external support in the form of problem solving hints and instructional information. This set of conditions requires the learner to adopt an active role in skill development and situates learning and extended practice in the context of real world problems. This instructional approach is in contrast to traditional, more passive skill training where the instruction amounts to telling students about a domain such as electronics rather than providing learning experiences for doing electronic problem solving.

A large research and development program is underway in the Air Force to train technicians for complex work environments in a manner that seeks to avoid these pitfalls. The Air Force Basic Job Skills (BJS) Research Program is examining the performance of technical experts in dozens of occupations to establish models of expertise as targets for training. Advances in knowledge engineering procedures such as those used in developing expert systems are being applied to specify in great detail the technical expert's strategies and supporting skill and knowledge bases. Of particular interest are dimensions of expert performance that cut across Air Force jobs and can thus be characterized as basic to expertise in complex work environments. In some sense these common dimensions can be viewed as modern day basic skills or the skills needed for a technologically advanced world. In addition, applications of artificial intelligence to instruction in the form of intelligent tutoring systems are being utilized to create the desired conditions for active, problem-oriented learning. In this paper, work done with over 15 experts in four related electronic and computer maintenance jobs will be highlighted to illustrate the "engineering" of expert knowledge. Also, a successful training study conducted with apprentice electronic technicians will be reported. In this study, the standard obstacles in complex skill training were satisfactorily overcome.