Software Engineering Institute

This article examines the transformative potential of generative artificial intelligence (Al) in redefining software engineering and acquisition practices. Distinguished by the ability to create new content from vast datasets, generative Al promises increased productivity and innovation that is particularly relevant for the Department of Defense (DoD). However, adopting generative Al poses both opportu­nities and challenges. This article delves into the nature of generative Al, focusing on large language models (LLMs) that produce text-based content and highlighting their application for tasks like code generation, document summarization and discrepancy analysis, and decision support. However, the potential pitfalls of Al, such as overfitting and biased outputs, necessitate robust validation methods and human oversight. We advocate integrating generative Al with human expertise to navigate the challenges and fully leverage its potential in software engineering and acquisition.