Analysis of Chatbots: History, Use Case, and Classification

Abstract

Chapter Preview

Top

Introduction

A chatbot, at its fundamental level, is a computer programme that simulates and handles human conversation (written or spoken), allowing users to engage with digital devices as if they were interacting with someone in real life. Chatbots can range from simple programmes that answer to a single line of text to sophisticated virtual assistants that learn and evolve to provide greater levels of personalisation as they collect and process data (Caldarini et al. 2022). These encounters can occur across a range of messaging platforms, websites, mobile apps, and even voice-enabled devices.

Whether you notice it or not, you have definitely interacted with a chatbot. For example, you may be searching a product on your personal device while a window appears and asks whether you need assistance. Perhaps you are on your way to a performance and you use your smartphone to request a taxi via chat. On the contrary, you may have employ voice commands to order a coffee from the nearby café and ended up getting a response notifying you of when it will be available and how much it would cost. Figure 1 depicts numerous scenarios in which you might encounter a chatbot.

Figure 1.

Resolution percentages by service issue type for users of Chatbots

(Gartner, 2023)

How Do Chatbots Act?

The first chatbots were essentially interactive frequently asked queries programmes that were built to respond to a limited set of frequently asked queries with pre-written replies. They often required users to select from simple terms and phrases to carry the conversation ahead because they were unable to interpret natural language. Such primitive traditional chatbots are incapable of processing complex inquiries or answering basic ones that creators have not expected.

In this section we present comprehensive detail about the key elements of each component of the overall architectural layout of any chatbot (Ashfaque, M. W. 2022). From Figure 2, we can have a basic idea about chatbot architecture. The User Interface, Natural Language Understanding (NLU), Dialogue Management, Backend, and Response Generation components make up the five primary parts of a generic chatbot design.

Figure 2.

General chatbot architecture

Chatbot algorithms evolved over time to be able to perform more complex rules-based programming along with natural language processing (Nithyanandam et al., 2021), allowing customers to express themselves in a conversational manner. This spawned a new kind of chatbot, one that is contextually aware and equipped with machine learning to continuously improve its ability to accurately understand and forecast requests as it is exposed to more and more human language.

Natural language understanding (NLU) is now used by modern AI chatbots to determine the meaning of open-ended input from humans, overcoming everything from typos to translation problems. The meaning is then mapped to the exact “intent” the user wants the chatbot to act on, and conversational AI is used to develop an appropriate answer. These AI technologies combine both machine learning and deep learning—distinct AI aspects with subtle differences—to build an increasingly precise knowledge base of questions and responses guided by user interactions. Because of recent advances in large language models (LLMs), this sophistication has resulted in higher customer satisfaction and more diverse chatbot applications.

Key Terms in this Chapter

Natural Language Understanding (NLU): It is a subfield of artificial intelligence (AI) and computational linguistics which seeks to enable computers to read and interpret human language in a meaningful and usable manner. It entails creating algorithms and models that enable machines to evaluate, comprehend, and extract understanding from natural language inputs such as voice or text.

Virtual Assistant: A virtual assistant (VA) is an application agent that may execute a variety of jobs or provide services for a user depending on human input, such as commands or questions, both written and verbal. Such technologies frequently include chatbot capabilities to replicate human conversation.

Artificial Intelligence: Artificial intelligence (AI) is the simulation of human intelligence capabilities and problem solving ability by technology, particularly computers. AI has specific applications such as expert systems, processing natural-language (NLP), speech recognition, and machine vision.

Generative AI: Generative AI is a kind of artificial intelligence based technology which can generate a variety of content, such as text, images, audio, and synthetic data. Traditional AI demonstrates its superiority in tasks that require logical reasoning, pattern identification, and rule-based decision making. Generative AI, on the other hand, excels at activities that need creativity, invention, and the ability to create new and original information.

Natural Language Processing (NLP): NLP refers to a computer program's capacity for understanding human language as written and spoken. It entails creating algorithms and models to help computers understand, interpret, and synthesize human language in a meaningful and usable manner. It is an interdisciplinary branch of computer science and linguistics.

Intelligent Personal Assistant: An intelligent personal assistant (IPA) is software meant to help people with simple tasks by presenting information in natural language. These virtual entities are designed with natural language processing, machine learning skills, and have access to a diverse set of data sources, allowing them to efficiently interpret and respond to user requests.

Large Language Models (LLM): LLMs are artificial intelligence systems that understand as well as generate human-like text. These models are often created with deep learning techniques, notably neural networks, and trained on massive volumes of text data. The word “large” refers to the neural network's size as well as the training data's scale.