What the software engineer should know about AI -- and vice versa

Abstract

Artificial Intelligence (AI) and Software Engineering (SE) have traditionally been largely independent disciplines. A common interest in developing computer systems is an area of overlap but fundamental differences in goals and objectives has resulted in little actual interaction. Limited success with chronic problems in both disciplines has forced practitioners in both fields to look beyond their local horizons in the hope of discovering new sources of impetus for further progress within their individual areas of interest. Software engineers are looking towards AI for both potential increase in software ‘power’ that AI techniques seems to promise, and exploitation of complexity-reduction strategies that AI has, of necessity, pioneered. AI practitioners, struggling with fragile demonstrations, would dearly like to import from SE tools, techniques and ‘know-how’ for the production of robust, reliable, and maintainable software systems. In this paper, I shall survey the range and scope of this two-way interaction between AI and SE.