1 Introduction to Generative AI on AWS

 

This chapter covers

  • Recent advancements in generative AI
  • Setting up your environment for developing with Amazon Bedrock
  • Understanding the generative AI solution lifecycle
  • Overview of Amazon Bedrock for building generative AI solutions

In the past two years, generative AI has changed the way individuals and organizations have perceived and made use of AI. Generative AI refers to models that can create new content—such as text, images, or videos—based on training data. While technologies like Long Short-Term Memory networks (LSTMs), Generative Adversarial Networks (GANs), and Variational Autoencoders (VAEs) have long been part of this field, recent advancements have significantly accelerated its progress. Within the scope of this book, generative AI focuses on models available on Bedrock, including Large Language Models (LLMs), Multimodal Language Models (MLLMs), and Diffusion Models. Consider the swift acceleration of the AI landscape since the release of models like Claude, enabling anyone to generate text for diverse use cases. Its capabilities have exceeded what most think AI can do—from generating new text to answering users’ questions to creating high-quality videos from a single line of text. Despite these breakthroughs, we only have a glimpse of its potential, as the field remains in the nascent stage of generative AI-driven innovation.

1.1 The Current Generative AI Landscape

1.1.1 Introduction to Foundational Models

1.1.2 Recent Advances in Generative AI

1.1.3 Industries Using Solutions with Amazon Bedrock

1.1.4 Tasks Performed by Generative AI

1.1.5 Ethics and Security

1.1.6 Technical Challenges

1.1.7 How Amazon Bedrock Approaches AI Challenges

1.2 Lifecycle of a Generative AI solution

1.2.1 Problem Definition and Scope

1.2.2 Data Collection and Preparation

1.2.3 Model Selection and Tuning

1.2.4 Training and Evaluation

1.2.5 Integration and Deployment

1.2.6 Monitoring, Security, and Compliance

1.2.7 Feedback and Improvement

1.3 Tech Setup for Using Generative AI on AWS

1.3.1 Python for Developing Generative AI Applications

1.3.2 LangChain for Generative AI Applications

1.3.3 Understanding How Amazon Bedrock Works

1.3.4 Understanding Pricing in Amazon Bedrock

1.4 Amazon Bedrock Capabilities

1.4.1 Forms of Generation

1.4.2 A One-stop solution

1.4.3 Creating Knowledge Bases

1.4.4 Fine-Tuning

1.4.5 Agents

1.4.6 Safeguards

1.4.7 Spaces for Experimentation

1.4.8 Working with other AWS Services

1.4.9 A Customer Service Chatbot Solution with Amazon Bedrock

1.5 Summary