Generative AI Learning Journey

A curated learning journey from Ruth Eneyi Ikwu from our Education team

Introduction to Generative AI

Generative AI refers to a subset of artificial intelligence that leverages machine learning techniques to generate data that is similar to the data it was trained on. The most common types of generative AI include Generative Adversarial Networks (GANs), Variational Autoencoders (VAEs), and Transformer-based models like GPT-3 and GPT-4.

  1. Generative Adversarial Networks (GANs): This technique consists of two parts: a generator network, which produces new data instances, and a discriminator network, which tries to distinguish between real and generated instances. The generator gets better at producing convincing output, and the discriminator gets better at detecting these instances, resulting in the generator producing high-quality data.

  2. Variational Autoencoders (VAEs): VAEs are a type of autoencoder that produces a continuous, structured output. They use a probabilistic approach to model the inherent uncertainty in predicting realistic outputs, and can generate entirely new data that has similar characteristics to the training data.

  3. Transformer-based models like GPT-3 and GPT-4: These models are designed to generate coherent and contextually relevant sentences by predicting subsequent words within a given text. They have been used to create impressive applications like writing articles, generating poetry, and even coding.

In each of these examples, the AI "learns" patterns and characteristics from its training data and uses that knowledge to generate new, original content. This makes Generative AI a powerful tool for a wide range of applications, including but not limited to creating art and music, designing new molecules for drug discovery, data augmentation, and many more.

Recommended resources (please note that each resource contains a link external to the Diverse AI website that will open in a new window)

• Learn The Basics of Generative AI (Medium Article by Reshma Abraham)

• Nvidia Simple Introduction to Generative AI (An Article from Nvidia)

• Google’s Generative AI Complete Course (Learning Path from Google)

• Generative Adversarial Networks (Course Specialisation from DeepLearning.ai)

• Building AI Powered Business Assistants without programming (Hand’s On Course from IBM)

• Fundamentals of Generative AI in Real World Business Applications (Short course from DeepLearning.ai)

• Generative AI For Everyone (Short Course from Andrew Ng)

Text Generative AI

Text Generative AI focuses on the creation of new, meaningful text based on the data it has learned from. Through complex algorithms and deep learning techniques, these AI models are able to understand patterns, context, and semantics in human language, enabling them to generate text that is remarkably coherent and contextually relevant.

The journey of Text Generative AI began with models like Recurrent Neural Networks (RNNs) that learned to process sequences of data, including text. This journey evolved further with Long Short-Term Memory Networks (LSTMs), designed to remember and maintain context over longer stretches of text. Today, the AI landscape is dominated by Transformer-based models, such as GPT-3 and GPT-4, which revolutionised text generation with their ability to better weigh the influence of different words in text generation, resulting in more human-like outputs.

Whether it's creating compelling articles, generating conversational responses for chatbots, drafting emails, or translating languages, Text Generative AI is transforming the way we interact with machines and how we create and consume content. As we delve deeper, prepare to discover how Text Generative AI models learn, how they function, their applications, and their potential to reshape our future.

Recommended resources (please note that each resource contains a link external to the Diverse AI website that will open in a new window)

Image Generative AI

Image Generative AI is a fascinating subset of artificial intelligence focused on creating new, original visual content. It employs advanced machine learning techniques to generate images that closely resemble the real-world data they've been trained on.

GAN’s and VAEs have been employed to generate incredibly realistic human faces, alter images (like turning a horse into a zebra), create artwork, and much more. However, the applications aren't confined to these areas. Image generative AI also has promising potential in industries like fashion, gaming, real estate, and healthcare, providing a tool for innovation and creativity.

Recommended resources (please note that each resource contains a link external to the Diverse AI website that will open in a new window)