Conversational AI

Andrew Freed

• Publisher: Manning
• Year: 2021
• ISBN:
  9781638351009
• Topic: Computer Science
• Subtopic: Computer Science General
• Index: Computer Science

"A thorough guide to the entire process of designing and implementing virtual assistants. Goes way beyond the technicalities." - Maxim Volgin, KLM Design, develop, and deploy human-like AI solutions that chat with your customers, solve their problems, and streamline your support services. In Conversational AI, you will learn how to: Pick the right AI assistant type and channel for your needs
Write dialog with intentional tone and specificity
Train your AI's classifier from the ground up
Create question-and-direct-response AI assistants
Design and optimize a process flow for web and voice
Test your assistant's accuracy and plan out improvements Conversational AI: Chatbots that work teaches you to create the kind of AI-enabled assistants that are revolutionizing the customer service industry. You'll learn to build effective conversational AI that can automate common inquiries and easily address your customers' most common problems. This engaging and entertaining book delivers the essential technical and creative skills for designing successful AI solutions, from coding process flows and training machine learning, to improving your written dialog. Purchase of the print book includes a free eBook in PDF, Kindle, and ePub formats from Manning Publications. About the technology
Create AI-driven chatbots and other intelligent agents that humans actually enjoy talking to! Adding intelligence to automated response systems saves time and money for you and your customers. Conversational AI systems excel at routine tasks such as answering common questions, classifying issues, and routing customers to the appropriate human staff. This book will show you how to build effective, production-ready AI assistants. About the book
Conversational AI is a guide to creating AI-driven voice and text agents for customer support and other conversational tasks. This practical and entertaining book combines design theory with techniques for building and training AI systems. In it, you'll learn how to find training data, assess performance, and write dialog that sounds human. You'll go from building simple chatbots to designing the voice assistant for a complete call center. What's inside Pick the right AI for your needs
Train your AI classifier
Create question-and-direct-response assistants
Design and optimize a process flowAbout the reader
For software developers. Examples use Watson Assistant and Python. About the author
Andrew R. Freed is a Master Inventor and Senior Technical Staff Member at IBM. He has worked in AI solutions since 2012.Table of Contents
PART 1 FOUNDATIONS
1 Introduction to conversational AI
2 Building your first conversational AI
PART 2 DESIGNING FOR SUCCESS
3 Designing effective processes
4 Designing effective dialogue
5 Building a successful AI assistant
PART 3 TRAINING AND TESTING
6 Training your assistant
7 How accurate is your assistant?
8 Testing your dialogue flows
PART 4 MAINTENANCE
9 Deployment and management
10 Improving your assistant
PART 5 ADVANCED/OPTIONAL TOPICS
11 Building your own classifier
12 Additional training for voice assistants

Part 1. Foundations

Have you ever used a chatbot? Do you turn the lights on and off by speaking to your phone? Does your email client sort your mail and filter out spam automatically? If so, you’ve interacted with an AI assistant. These are just a few examples of the ways that AI assistant technology is used.

Conversational AI and other AI assistants are transforming the way businesses interact with their customers. These assistants bring self-service power to users, helping them get what they want, when they want it, the way they want it. Businesses around the world are saving millions of dollars by using AI and making their customers happier at the same time.

In chapter 1, you’ll learn about three types of AI assistants and when they are appropriate to use. You’ll also learn about the most common AI assistant use cases and platforms. In chapter 2, you’ll see how a conversational AI is assembled from three core building blocks: intents, entities, and dialogue responses.

When you complete this part, you’ll be ready to start designing your AI assistant solution.

  

1 Introduction to conversational AI

This chapter covers

• Listing the types of AI assistants and their platforms
• Classifying AI assistants
• Recognizing AI assistants that you already interact with
• Differentiating questions needing a simple response versus a process flow
• Describing how email clients learn from your actions

AI assistants are an exciting new technology that is being used in an increasing number of places and use cases. The odds are good that you have interacted with an AI assistant. From the assistants on our phones (“Hey Siri”) to automated customer service agents to email filtering systems, AI assistant technology is widespread and growing in use.

In this book, you will learn how and why AI assistants work. You will learn when AI assistant technology is appropriate and how to apply the technology effectively. You will also learn how to develop, train, test, and improve your AI assistant. Finally, you will learn about advanced topics, including enabling a voice channel for your AI assistant and deeply analyzing your assistant’s performance.

Who is this book for? Who is this book not for?

This book is written for someone who is interested in developing AI assistants. We will start with broad coverage of multiple assistant types and how they are used; then we will take a deep dive into all aspects of creating a conversational AI assistant, including design, development, training, testing, and measurement.

If you have already developed several assistants, you’ll probably want to skip to specific chapters later in the book. If you are not a developer, you can read the first couple of chapters and then skim or skip the rest.

1.1 Introduction to AI assistants and their platforms

Why are AI assistants popular? Let’s examine one industry that frequently uses AI assistants: customer service. A new call or case at a customer service center averages between 8 and 45 minutes, depending on the product or service type. Customer service call centers spend up to $4,000 for every agent they hire, and even more money in training costs, while experiencing employee turnover of 30–45%. This situation leads to an estimated annual loss of $62 billion in sales annually in the United States alone (http://mng.bz/G6rR). AI assistants can help with these problems and more.

You’ve probably had several recent interactions with different kinds of AI assistants:

• Chat interfaces for customer service and guided buying
• Smartphones and connected homes that are controlled by voice (“Alexa, turn on the light!”)
• Email software that automatically sorts your mail into folders such as Important and Spam

AI assistants are pervasive and used in a wide variety of ways. How many of the AI assistants in table 1.1 have you interacted with? The rest of this section dives into the mainstay platforms and AI assistant types in this ecosystem.

Table 1.1 Examples of AI assistants and their platforms

AI assistant type	Uses	Skills you’ll need to build one	Technology focus	Example platforms
Conversational AI (sometimes called a chatbot)	Customer service Guided buying experience New employee training	Designing and coding process flows with one or many steps Using conversational state to determine the next step Writing dialogue Classifying utterances into intents Extracting entities from utterances to support the intent Writing code to call external application programming interfaces (APIs)	Conversation and dialogue flow	IBM Watson Assistant Microsoft Azure Bot Service Rasa
Command interpreter	Natural language or voice interface to devices	Classifying statements into commands Extracting supporting parameters from a command statement Writing code to call external APIs	Classification and calling APIs	Apple Siri Amazon Alexa
Event classifier	Sorting email into folders Routing messages (such as emails) to an appropriate handler	Classifying messages based on message content and metadata Extracting many entities that support or augment the classification	Classification and entity identification	Google Gmail Natural Language Understanding services

A wide variety of platforms can help you build AI assistants. Most of the platforms are usable in a variety of AI assistant scenarios. You should consider a few things when you choose a platform:

• Ease of use—Some platforms are intended for business users, and some are for software developers.
• APIs and integration—Does the platform expose APIs or have prebuilt integrations to third-party interfaces and tools? An AI assistant is usually integrated into a larger solution.
• Run-time environment—Some platfor...

Table of contents

• inside front cover
• Conversational AI
• Copyright
• dedication
• brief contents
• contents
• front matter
• Part 1. Foundations
• 1 Introduction to conversational AI
• 2 Building your first conversational AI
• Part 2. Designing for success
• 3 Designing effective processes
• 4 Designing effective dialogue
• 5 Building a successful AI assistant
• Part 3. Training and testing
• 6 Training your assistant
• 7 How accurate is your assistant?
• 8 Testing your dialogue flows
• Part 4. Maintenance
• 9 Deployment and management
• 10 Improving your assistant
• Part 5. Advanced/optional topics
• 11 Building your own classifier
• 12 Additional training for voice assistants
• Appendix. Glossary of terms and abbreviations
• index
• inside back cover