What Is Generative AI?

Over the past decade, deep learning has evolved massively to process and generate unstructured data like text, images, and video. These advanced AI models have gained popularity in various industries, and include large language models (LLMs). There is currently a significant amount of fanfare in both the media and the industry surrounding AI, and there’s a fair case to be made that Artificial Intelligence (AI), with these advancements, is about to have a wide-ranging and major impact on businesses, societies, and individuals alike. This is driven by numerous factors, including advancements in technology, high-profile applications, and the potential for transformative impacts across multiple sectors.

In this chapter, we’ll explore generative models and their basics. We’ll provide an overview of the technical concepts and training approaches that power these models’ ability to produce novel content. While we won’t be diving deep into generative models for sound or video, we aim to convey a high-level understanding of how techniques like neural networks, large datasets, and computational scale enable generative models to reach new capabilities in text and image generation. The goal is to demystify the underlying magic that allows these models to generate remarkably human-like content across various domains. With this foundation, readers will be better prepared to consider both the opportunities and challenges posed by this rapidly advancing technology.

We’ll follow this structure:

• Introducing generative AI
• Understanding LLMs
• Model development
• What are text-to-image models?
• What can AI do in other domains?

Let’s start from the beginning – with the terminology!