In this article we’ll cover chatbot fundamentals, including what a chatbot is, how it works and why it’s important.
So, if you’re just getting started with chatbots, or want to strengthen your knowledge, this chapter is for you.
A chatbot is a computer program that allows humans to interact with technology using a variety of input methods such as voice, text, gesture and touch, 24/7 365.
For several years chatbots were typically used in customer service environments but are now being used in a variety of other roles within enterprises to improve customer experience and business efficiencies.
Known by a variety of different names such as a conversational AI bot, AI assistant, intelligent virtual assistant, virtual customer assistant, digital assistant, conversational agent, virtual agent, conversational interface and more, chatbots are growing in popularity.
But just as chatbots have a variety of different names, they also have varying degrees of intelligence.
A basic chatbot might be little more than a front-end solution for answering standard FAQs.
Chatbots built using some of the bot frameworks currently available may offer slightly more advanced features like slot filling or other simple transactional capability, such as taking pizza orders.
But, it’s only advanced conversational AI chatbots that have the intelligence and capability to deliver the sophisticated chatbot experience most enterprises are looking to deploy.
For the purpose of this guide, all types of automated conversational interfaces are referred to as chatbots.
Smartphones, wearables and the Internet of things (IoT) have changed the technology landscape in recent years. As digital artefacts got smaller, the computing power inside has become greater.
But mobile apps and data-heavy activities don’t go hand in hand. Wading through complicated menus isn’t the fast and seamless user experience businesses need to deliver today.
In addition, consumers are no longer content to be restricted by the communication methods chosen by an organization. They want to interface with technology across a wide number of channels.
Chatbots offer a way to solve these issues by allowing customers to simply ask for whatever they need, across multiple channels, wherever they are, night or day.
On a simple level, a human interacts with a chatbot.
If voice is used, the chatbot first turns the voice data input into text (using Automatic Speech Recognition (ASR) technology). Text only chatbots such as text-based messaging services skip this step.
The chatbot then analyses the text input, considers the best response and delivers that back to the user. The chatbot’s reply output may be delivered in any number of ways such as written text, voice via Text to Speech (TTS) tools, or perhaps by completing a task.
It’s worth noting that, understanding humans isn’t easy for a machine. The subtle and nuanced way humans communicate is a very complex task to recreate artificially, which is why chatbots use several natural language principles:
Natural Language Processing (NLP)
Natural Language Processing is used to split the user input into sentences and words. It also standardizes the text through a series of techniques, for example, converting it all to lowercase or correcting spelling mistakes before determining if the word is an adjective or verb — it’s at this stage where other factors such as sentiment are also considered.
Natural Language Understanding (NLU)
Natural Language Understanding helps the chatbot understand what the user said using both general and domain specific language objects such as lexicons, synonyms and themes. These are then used in conjunction with algorithms or rules to construct dialogue flows that tell the chatbot how to respond.
Natural Language Generation (NLG)
Delivering a meaningful, personalized experience beyond pre-scripted responses requires natural language generation. This enables the chatbot to interrogate data repositories, including integrated back-end systems and third-party databases, and to use that information in creating a response.
Conversational AI technology takes NLP and NLU to the next level. It allows enterprises to create advanced dialogue systems that utilize memory, personal preferences and contextual understanding to deliver a realistic and engaging natural language interface.
Chatbots can trace their history back decades, but it wasn’t until internet usage became more mainstream that the chatbots as we recognize them today, started to be used to support customer service functions.
Here’s a breakdown of some of the more prominent moments defined in chatbot history:
Turing Test, 1950
The Turing Test asks the question of whether machines can think, and was asked in 1950 by Alan Turing in his 1950 landmark paper, “Computing Machinery and Intelligence”. In the paper, Turing proposed a test where an interrogator had to determine which player was a human and which a machine through a series of written questions.
Despite criticisms and flaws, the test is still performed regularly today.
In 1964, MIT computer scientist Joseph Weizenbaum started development on ELIZA, what would turn out to be the first machine capable of speech using natural language processing.
Symbolically named after Eliza Doolittle in George Bernard Shaw’s Pygmalion, ELIZA was able to fool many people into believing they were talking to a human simply by substituting their own words into scripts and feeding them back to users to maintain the conversation.
By the early 1970s, psychiatrist Kenneth Colby had taken the principles behind ELIZA a step further. With the introduction of PARRY, Colby adopted more of a conversational chatbot strategy than ELIZA using a model of someone with paranoid schizophrenia to help increase believability in the responses. In 1973 a conversation was set up between ELIZA and Parry.
RACTER, the “artificially insane” raconteur, was written by William Chamberlain and Thomas Etter. It was reportedly said that the book ‘The Policeman’s Beard’ was written by the Chatbot Racter. However, Racter was never released publicly.
Jabberwacky is a chatterbot created by British programmer Rollo Carpenter. It was one of the earliest attempts at creating AI through human interaction. The chatbot was designed to “simulate natural human chat in an interesting, entertaining and humorous manner”.
Loebner Prize, 1990
The Loebner Prize was launched in 1990 by Hugh Loebner. It takes the format of a standard Turing Test with judges awarding the most human-like computer program.
Dr. Sbaitso, 1991
Dr. Sbaitso was a computerized psychologist chatbot with a digital voice designed to speak to you. It was an artificial intelligence speech synthesis development, created by Creative Labs meant to show off the sound card’s then-impressive range of digitized voices.
A.L.I.C.E. (Artificial Linguistic Internet Computer Entity) also referred to as Alicebot, or simply Alice, is a natural language processing chatterbot first developed in 1995, who has won the Loebner three times. Alice was inspired by the ELIZA program.
Elbot is the cheeky chatbot who uses sarcasm and wit, along with a healthy dose of irony and his own artificial intelligence to entertain humans. Elbot was created by Fred Roberts and Artificial Solutions. In 2008 Elbot was close to achieving the 30% traditionally required to consider that a program has passed the Turing Test.
The Smarterchild chatbot was developed by ActiveBuddy Inc. by Robert Hoffer, Timothy Kay and Peter Levitan. It was available on AOL Instant Messenger MSN Messaging networks. The chatbot offered fun personalized conversation and was considered a precursor to Apple’s Siri and Samsung’s S Voice.
Mitsuku is a chatbot created from AIML technology by Steve Worswick. It’s a five-time Loebner Prize winner (in 2013, 2016, 2017, 2018, 2019). Mitsuku claims to be a teenage female chatbot from Leeds, England. Her intelligence includes the ability to reason with specific objects, she can play games and do magic.
IBM Watson, 2006
Named after IBM’s first CEO, Thomas, J. Watson, Watson was originally developed to compete on the American TV program, ‘Jeopardy!’, where it defeated two of the former champions in 2011. Watson has since transitioned to using natural language processing and machine learning to reveal insights from large amounts of data.
Siri first came to the public’s attention in February 2010 when it was launched as a new iPhone app. Apple subsequently bought the company and integrated the voice assistant into the iPhone 4S at its release in October 2011, bringing voice applications into the mainstream consumer market for good.
Google Now, 2012
Google Now was developed by Google, created specifically for the Google Search Mobile App. It uses a natural language user interface to answer questions, make recommendations, and perform actions by passing on requests to a set of web services.
Siri remained perhaps the most famous of mobile voice assistants until Amazon launched Alexa. Already familiar with giving commands to their phone, Alexa caught consumers imagination and launched the now-immense market for smart home speakers.
Cortana is an intelligent personal assistant that was developed by Microsoft. Cortana recognizes natural voice commands, can set reminders and answer questions using the Bing search engine.
Bots for Messenger: Facebook Chatbots, 2016
With Facebook’s launch of their messaging platform, they became the leading program for chatbots. In 2018 there were more than 300,000 active chatbots on Facebook’s Messenger platform.
Tay was a chatbot created by Microsoft to mimic the speech and habits of a teenage American girl. The chatbot caused controversy and was shut down only 16 hours after launch, when it began to post offensive tweets and became increasingly paranoid.
Woebot developed by Woebot Labs is an AI-enabled therapy chatbot designed to help users learn about their emotions with “intelligent mood tracking.”
2020 and Beyond
Expect to see enterprises planning for an intranet of conversational AI applications that can work together seamlessly, sharing information.