A Historical Analysis of Chatbots from Eliza to Google Bard

The increasing popularity of Chatbots brings forth many opportunities for advancement in the field of conversational Artificial Intelligence (AI), expanding to a wide range of fields including customer services, education, marketing, health care, entertainment, and many more. Nowadays, ChatGPT and Google Bard are considered the leading Chatbots because of their ability to converse with humans in a more advanced and modern way. This paper provides a comprehensive review of the history, machine learning (ML) approaches, large language models (LLMs), and languages behind Chatbots. It begins by tracing the history of Chatbots from their early beginnings to the present day, highlighting the key developments along the way; followed by a discussion of four main ML approaches (rule-based, retrieval-based, generative, and hybrid) used in Chatbots with their architectures and some examples. After that, generative AI-based LLMs approaches for ChatGPT and Google Bard, which follow GPT-3.5 and LaMDA, respectively, are discussed in detail. The review concludes by highlighting that both ChatGPT and Google Bard have been trained on huge datasets and generate output as per instructions given by the user. However, both models have some limitations that need to be addressed to improve their usefulness in terms of reliability and capabilities.