Desarrollo de aplicaciones a la medida con apoyo de modelos GPT

Abstract

Based on the need to improve productivity in custom application development, experts at Dirección General de Cómputo y de Tecnologías de Información y Comunicación explored different methodologies, techniques and tools. It was reviewed that GPT models contribute to improve the learning curve of custom application developers in programming frameworks and languages, thereby reducing the development time of the various software components that make up custom applications. This report explains the elements that can be integrated into the design of the prompt or request and how the prompt can be refined so that GPT models provide better responses that are easier to integrate into the application under development and less complex in their logic. The complexity of the components was taken into account using the Laravel Sail Insights tool. It was concluded that applying generative Artificial Intelligence GPT models reduces the construction time of functions, components and software components, as well as the number of issues to be solved. Furthermore, issues detected by the software quality assurance team are dealt with in less time. An experienced developer can design the best prompts to elicit the best responses from GPT models and analyze the results to decide whether to incorporate the generated code, while also taking into account the ethical framework of copyright issues.