Chatbots are increasingly playing a role in e-commerce and the services sector. Their use has many advantages in terms of economy, efficiency and dependable service. This article explores how chatbots work.
Many experts have described 2016 as ‘The year of the chatbots’. Thousands of chatbots are today helping businesses to improve customer service, sell more and increase their earnings. These chatbots are being used to carry out a number of business tasks in the e-commerce, insurance, banking, healthcare, finance, legal, telecom, logistics, retail, auto, travel, sports, entertainment and other fields. One report predicts that more than 85 per cent of customer interactions will be managed without humans by 2020.
A chatbot is a computer program or artificial intelligence that conducts a conversation using auditory or textual methods. Such programs are often designed to convincingly simulate how a human would behave as a conversational partner, thereby passing the Turing test.
Today, chatbots are intelligent assistants for human beings. They can perform several human tasks like managing calendars, making reservations, booking tickets, placing food orders, etc. With smart homes, voice assistants (like Amazon Alexa and Google Home) will soon be able to perform many more actions. In fact, between 2016 and 2021, the chatbot market is expected to grow at a compound annual growth rate (CAGR) of 35.2 per cent.
Platforms for building chatbots
In 1994, Michael Mauldin coined the term ‘Chatter Bot’ to describe conversational programs or chatting robots, and this got shortened to chatbot. Today, most chatbots are accessed using voice assistants like Google Assistant and Amazon Alexa or through messaging apps such as Facebook Messenger or WeChat.
Currently, there are a lot of platforms for building chatbots. Though not necessarily open source, these are being built by developers using open source technologies. IBM Watson, Microsoft Bot framework, LUIS, Wit.ai, Api.ai and Chat Fuel are some of the best known platforms for building chatbots. Let’s take a look at some of these platforms.
IBM Watson: It’s estimated that nearly 61 per cent of the chatbot based business solutions are today being developed on the IBM Watson bot-building platform. One of Watson’s most important components is conversation services. It is built on a neural network, understands intentions, interprets dialogues, supports English and Japanese, and provides tools like the SDKs for Java, Python, iOS and Unity. IBM offers free, standard premium plans.
Microsoft Bot Framework: According to one study, 41 per cent of businesses prefer the Microsoft Bot Framework. It has its own Bot Builder SDK that includes the SDK for .NET and Node.js. The entire system consists of three parts – the bot connector, the developer portal and the bot directory. This framework is open source, available to all on GitHub, and supports automatic translation to more than 30 languages. It incorporates LUIS (Language Understanding Intelligent Service) for natural language understanding, Cortana for voice and Bing APIs for search.
Dialog Flow: This is a mature chatbot-creation platform, perfectly suited to build intent-based chatbots. It was previously known as Api.ai. It currently has SDKs for Android, iOS, Cordora, Unity, Xamarin, HTML, JavaScript, Node.js, Epson Moverio, C++, Python, Ruby, PHP, Java, etc. A total of 14 languages are supported. It also comes with training modules.
Amazon Lex: This is the background technology used in the NLP/NLU engine for Amazon Alexa.
Many Chat: Many Chat lets you create a Facebook Messenger bot for marketing, sales and support. It’s easy to use and free.
Semantic Machines: The features of this platform include a conversation engine, speech synthesis, deep learning, reinforcement learning, speech recognition, semantic intent extraction and natural learning generation (NLG) technology. It goes beyond just understanding conversations. This platform is language-independent.
Chat Fuel: More than 360,000 chatbots have been created using Chat Fuel, serving more than 17 million users globally. Many plugins for it have been developed. Some of them are: Google Search, Bing Search, JSON API, RSS import, subscribe plugin, Digest, Zapier, user input, live chat, etc. Chat Fuel supports about 50 languages and is free.
Chatter Bot: This is a Python library that makes it possible to generate responses based on a collection of known conversations. The Chatter bot language is independent.
Keeping service in mind
Have you ever used a customer support live chat service? You may at that time have got the impression that you are chatting with a person rather than a robot.
Whether you love chatbots or hate them, they are here to stay. They are becoming extremely popular in today’s Web context due to the dramatic advances in machine learning, natural language processing, etc. Today’s chatbots are friendlier, smarter, more responsive and more informative. And we may see even more features added in the coming years. Today one can make one’s own chatbot that can be used on Facebook Messenger.
Briefly described below are some companies that use chatbots to provide better customer service.
Casper – Helping sleep disorders: Are you facing sleepless nights? Are you feeling lonely? Enter Casper’s Insomnobot 3000, a conversational agent that offers relief to insomniacs.
Disney – A chatbot to solve crimes with fictional characters: Popular characters from animation movies are now becoming chatbots. Disney has launched one of its characters from the hit ‘Zootopia’ as a chatbot on Facebook Messenger. The conversational agent is Lieutenant Judy Hopps, the much admired rabbit from the movie.
Med What – Makes medical diagnosis faster: If you are bored with medical websites or self-diagnosing yourself with life-threatening diseases, the Med What chatbot is of help. It makes medical diagnosis faster, easier and more transparent for both patients and doctors.
Roof Ai – Generating and assigning leads automatically: Are you a marketing executive? Then you may know the importance of route assignment. Roof Ai may help real estate marketers to find potential customers for their houses and apartments. The bot identifies potential leads via Facebook and responds almost instantaneously in friendly, helpful and conversational tones that closely resemble that of a real person.
Types of chatbots
Chatbots are available in all shapes and sizes with varying capabilities. Basic chatbots are enough for most of the use cases. For some requirements, advanced chatbots may be required.
There are basically three types of chatbots, as listed below.
Menu/button based chatbots: These are the most basic type of chatbots available today. They usually come with decision tree hierarchies presented in the form of buttons. Similar to the automated phone menus, these chatbots require the user to make several selections to dig deeper towards the ultimate answer.
Keyword recognition based chatbots: These chatbots can attend to what the user types and respond appropriately, or at least try to. These chatbots use customisable keywords and AI to give solutions to customers’problems.
Today, a hybrid of key word recognition based and menu/ button based chatbots are also available in the market. This sophisticated technology, which is based on the principles of AI and data mining, can detect and interpret the problems of customers pretty accurately.
Contextual chatbots: These are far more advanced, incorporating both machine learning (ML) and artificial intelligence (AI) .They remember conversations with specific users, and learn and become smarter with time. These bots come with self-improvement characteristics.
Let’s suppose a customer uses a contextual chatbot to order pizza. The chatbot will store the data from each conversation – details like what the user likes to order frequently. Next time, when the customer interacts with this chatbot to order pizza, it merely asks. “Do you want to repeat this order?” As a result, instead of having to respond to several questions, the user has to just answer with a “Yes,” and the pizza is ready to be despatched.
What’s trending?
The bot market is slowly maturing, and companies are trying to increase the features and capacity of chatbots.
Facebook is presently using natural language processing (NLP) technology. This is advanced enough to understand all kinds of user requests. But the new trend is natural language understanding (NLU). Many companies are also investing a lot of money in developing new technologies like reinforcement learning (RL). All this will make chatbots more effective.
Many people now feel that voice is the best way to converse with chatbots. In the age of Amazon Echo, Google Home and Apple Home, voice assistants are becoming common. User interface elements like images, interactive buttons and message menus make chatbots more effective, resulting in bot platforms like Slack, Telegram, etc.
Labour efficiency, cost savings and enterprise automation are the main benefits of chatbots. Today, an employee need not spend hours on routine office chores like password resets, printer issues, technical support, etc. There are IT Helpdesk bots for that! In short, AI chatbots can take employee experiences in the workplace to a new level in the future.
The obstacles
There are many challenges while dealing with chatbots. Different people have their own ways of typing messages. Some use short sentences and some long sentences. In such cases, to understand user intention is really a challenge for the chatbot. Further, these bots need to have robust security. So before making a chatbot live, it is necessary to test security protocols to make sure that there are no gaps in its defences.
Chatbots need to be useful, reliable, accurate and seem trustworthy to customers. Also, sophisticated knowledge about machine learning coding, NLP software and other technical knowhow is required for developing a chatbot.
The future
Brand-building is one of the biggest challenges for a business. So a chatbot should add something to a marketing site to build up the brand image of an enterprise. It should have rich analytic programs to analyse the most common requests by customers. Technology is changing frighteningly fast. So it is necessary to provide future-proof chatbots.
Chatbots can be game changers in the marketing, sales and services sectors. By 2020, 80 per cent of people’s needs are expected to be met by chatbots instead of expensive apps. Chatbots do not need any installing or downloading requirements. 24×7 real-time communication by chatbots can definitely strengthen a company’s customer support. Today, chatbots can mine information from various digital data channels using sophisticated data mining technology and provide an effective solution to your problems.
Chatbots are being widely used across the US, where several startups and enterprises are entering the business. Usage of chatbots is also increasing in Asia. The Chinese prefer to integrate the WeChat bot into retail, daily life and social media usage. In Japan, prompt and polite customer service is important to businesses and consumers. Here, chatbots are enabling customer loyalty programs and services. In Taiwan, chatbots are being used extensively for banking and financial services. When we speak about India, there is a lot of diversity—in needs, culture, infrastructure and usage. So any chatbot platform has to be able to encompass all of these smoothly.
There are thousands of chatbots in India as of now. Big brands like HDFC, Yatra, etc, are introducing chatbots into their websites. India has traditionally been a late adopter of advanced technologies. However, when it comes to chatbots, India is a key player. There were more than a hundred chatbot startups in India as of 2017.
With chatbots playing a vital role in today’s IT world, researchers are under pressure to develop reliable and economical designs. The need of the hour is to be able to predict the behaviour of chatbots to a very high degree, in order to ensure reliability. This has led to advanced chatbots with increased data flow, which are set to become an industry default. Researchers are working on creating chatbots capable of collecting, processing and analysing a huge amount of data, helping users to get useful and effective solutions to their problems.
