Bridging Data and AIOps for Future AI Advancements with Human-in-the-Loop. The AI-DAPT Concept

Abstract

The transition of artificial intelligence (AI) from research to deployment has underscored the critical importance of leveraging data effectively in developing and evaluating AI models. Despite their pivotal role in determining performance, fairness, and robustness, data are often undervalued in AI research, lacking a data-centric focus. In response, the AI-DAPT project pioneers a data-centric approach that aligns with AI methodologies to address the pressing need for reliable and trustworthy AI systems. This paper presents the AI-DAPT project and concept, highlighting its innovative solutions. Through advanced techniques like synthetic data generation and hybrid science-guided Machine Learning (ML), AI-DAPT aims to mitigate vulnerabilities in conventional AI paradigms. The developed solutions will be demonstrated in healthcare, robotics, energy, and manufacturing scenarios, hence navigating real-world complexities.