Generative Insights Unveiling AI's Evolution and Algorithms

Abstract

Artificial intelligence (AI) has the capability for machines to learn from experience, change their inputs and perform actions as if they were human. To begin with, AI researchers focused on primitive algorithms that have predefined rules. There are also a number of shortcomings when it comes to these algorithms like interpretability problems, inadequate data sources, computation resource, data description and quality, ethical consideration, overfitting and underfitting computational burden, data collection and bias, high probability of error, lack of enough trained instances no causality as well as reproducibility issues among others. This chapter will guide you through understanding of Generative AI by discussing fundamental algorithms and models used in powering this game-changing technology. In this, it investigates into the basics by using some generative algorithms like probability-based models, VAEs, GANs and autoregressive models.