Leveraging Generative AI Concepts with Uncertainty of Information Concept

The rapidly growing area of artificial intelligence (AI) that is used to create new content using machine learning is referred to as generative AI. The neural network transformer architecture and unsupervised as well as supervised learning have created the backbone of these generative models. These transformer-based large foundation models that can handle different modalities of data are the core building blocks of the generative AI concept. One requirement is data, ideally large, quality, unbiased data sets. With this data, these generative models learn patterns to form new content with characteristics of the original data. The Uncertainty of Information (UoI) concept is motivated by the fact that data is not ideal, is not perfect. Moreover, UoI considers the challenge of both capturing and communicating uncertainty associated with data. Given that generative AI, like other techniques, requires significant data, tying together these two concepts may uncover new capabilities. This paper will explore how these concepts can be leveraged, possible benefits, and challenges.