{"@context":"https://schema.org","@type":"ScholarlyArticle","headline":"Artificial Intelligence in Software Engineering","author":[{"@type":"Person","name":"D. Partridge"}],"datePublished":"2002","isPartOf":{"@type":"Periodical","name":"Encyclopedia of Software Engineering"},"identifier":"https://doi.org/10.1002/0471028959.SOF013","description":"AbstractArtificial intelligence (AI) and software engineering (SE) have, to a large extent, developed and matured independently, but recent years (at the time of writing) have brought significant cross‐fertilisation and the consequent developments exhibit the expected hybrid vigor. These two fields have a lot in cornmon—both aim to implement software systems as the prime goal. They also have many differences, especially when viewed historically. AI was, and perhaps still is, a domain of researchers whose aims were to reproduce aspects of human intelligence on computers and thereby learn something about the nature of computational intelligence per se, or about the embodiment of this intriguing phenomenon in humans. Software engineers, on the other hand, were preoccupied with developing a science and technology for efficiently instantiating quite well‐defined problems, robustly and reliably, in computer systems. The activities of both groups centered around programs and programming, but there the similarity ended. A successful outcome was measured rather differently within these two fields: (1) a reliable (perhaps verified) implementation of a well‐defined function for the software engineer and (2) a selection of “intelligent” behaviors from a concrete instantiation of a complex, incomplete, and ill‐defined theory for the AI researcher. So it is not too surprising that these two fields developed somewhat independently of each other.It is useful to outline a framework within which one can consider the full spectrum of impacts of AI on software engineering. There are three (or possibly four) major classes of interaction of AI and SE: software support environments, AI tools and techniques in conventional software, and use of conventional software technology in AI systems. The fourth area is methodological considerations. The impact of AI on software engineering, from a methodological perspective, has been to promote recognition of the inadequacies of a linearized specify‐and‐verify approach to software production. Instead of AI workers learning to “do things properly,” to specify their problems completely before designing solutions (although this has happened to some extent as well), the impact has been that software engineers recognize that interative software development can be a legitimate approach.","url":"https://innscience.ca/es/community/literature/cbd1c8b5-74be-452b-b55f-f9d2d14ad01d"}