The TURING Test Approach to Validation

This paper is based on the insight that intelligent system validation is a complex social process involving issues ranging from mathematics and software engineering to cognitive psychology and social sciences. To cope with the intrinsic complexity of the problem, there is developed a so-called \Tur-ing Test Methodology" for intelligent system validation. This is substantially distinguished from the perspective underlying the Loebner Prize business. There is an urgent need to provide a rm basis for integrating very diverse disciplines towards a comprehensive approach to validation. For this reason, much emphasis is put on the conceptual fundamentals. First, several concepts of veriication and validation are surveyed. The favoured concepts are lucidly characterized by the words that ver-iication guarantees to build the system right whereas validation deals with building the right system. Second, the thought-experiment called the Turing test is critically inspected. It turns out that, although this approach may not be suu-cient to reveal a system's intelligence, it provides a suitable methodological background to certify a system's validity. The principles of the validation approach are surveyed. Finally, further steps beyond the present elementary approach are outlined. On the one hand, there are systematically presented a few important generalizations. On the other hand, an implementation of the basic Turing test scenario is sketched. In the coming phase of their research and development project, the authors will invoke the methodology developed and outlined here for prototypical system validation. However, this is beyond the present paper.