PROPUESTA DE UN MARCO TEÓRICO PARA LA EVALUACIÓN DE LA CALIDAD DE LOS DATOS EN LA INTELIGENCIA ARTIFICIAL

Abstract

This article proposes a theoretical framework for data quality assessment in artificial intelligence (AI), addressing the lack of uniform standards in this crucial field. The framework focuses on standardized criteria to improve the consistency and reliability of AI systems, identifying specific areas of improvement in data collection and processing. The introduction highlights the importance of data quality in AI, and how it affects the accuracy and effectiveness of algorithms. The lack of clarity in the existing methods to evaluate this quality is highlighted, which motivates the proposal of the theoretical framework. The theoretical framework is based on a systematic review of the literature and qualitative analysis of case studies in sectors such as health and finance. Critical elements for data quality are identified: accuracy, integrity, relevance, timeliness and consistency. The proposed framework includes guidelines for applying these criteria in different phases of the life cycle of an AI project. The results demonstrate that the implementation of the framework improves the quality of the data and, therefore, the effectiveness of the AI ​​systems. Feedback from AI experts validates the relevance and usefulness of the framework. The discussion focuses on the importance of the framework, its alignment with existing literature, and the need for an ethical and responsible approach in AI. Future directions for research are suggested, highlighting the need to adapt and validate the framework in different subfields of AI. In conclusion, this study offers a robust and practical theoretical framework for data quality assessment in AI, with the potential to significantly improve the accuracy and reliability of these systems.