Software Engineering Institute

In this report we discuss in detail pragmatic problems posed by the integration of AI with conventional software engineering, and within the framework of current ADA technology. A major objective of this work has been to begin to bridge the gap between the ADA and AI software cultures. The report summarizes survey results from the Association for Computing Machinery (ACM) Special Interest Group for ADA (SIGADA) AI Working Group (AIWG), highlighting lessons learned and sample applications. An interesting observation is that a large percentage of AI code is procedural by nature and that better productivity rates are achieved by using ADA; also, efficiency is much better when compared to traditional AI languages. Although we show favorable results on the use of ADA technology for the implementation of AI software, a total integration remains difficult at the conceptual level. There are some impediments to a completely satisfactory solution, but only a few restrictions are more intrinsically related to the current ADA standard (ADA83); these, however, are being dealt with in the next revision known as ADA9X.