The increasing use of Chatbots shows that machines are constantly evolving to understand better and fulfil the information needs of humans.
By Prakash Kumar
These days when we open websites of banks, insurance companies, e-commerce companies, airlines, hotel chains, governments, restaurants etc., we see a tiny creature popping up from the right or left corner of our computer eager to help us. They come with very catchy names and answer simple questions, get bills/invoices, or do simple chores that a call centre agent used to do in the past. They are Chatbot, a word derived from “chat robot”.
What exactly is a Chatbot? A chatbot is a computer program that simulates human conversation or a chat, either via voice or text communication in natural language through messaging applications, websites, mobile apps, or the telephone. They are used for various purposes, including customer service, information dissemination, request routing, and information gathering alongside classic helpdesk. They work 24X7 without getting tired or needing a break, and hence more and more organizations are going for them. What started as a trickle in 2017 has made good inroads in the service industry.
How does a Chatbot work? There are two different tasks at the core of a Chatbot, user request analysis and returning the response. It analyses the user’s request (inputs) to identify the user intent and extract relevant entities. This process may look simple, but things are quite complex in practice. Identifying the user’s intent and extracting data and relevant entities contained in the user’s request is the first condition and most relevant step. If Chatbot cannot correctly understand the user’s request, it will not provide the correct answer. Identification of intent differentiates between different types of Chatbot, which use different technologies to achieve this. If Chatbot has not understood the question, it asks a question that helps it understand the user’s request correctly.
Once the user’s intent has been identified, the Chatbot provides the most appropriate response for the user’s request. The answer may be a generic and predefined text or a text retrieved from a knowledge base that contains different answers or the result of an action that the Chatbot performed by interacting with one or more backend applications.
The simplest Chabot scans for general keywords in the input provided by the user and generates responses using common phrases obtained from an associated library or database of questions and answers. These are generally first-generation Chatbots, which are rule-based. They recognise clue words or phrases in the input and provide an output of the corresponding pre-prepared or pre-programmed responses that can move the conversation forward in a meaningful way. Thus, an illusion of understanding is generated, even though the processing involved has been superficial. Their ability to answer questions is limited and creates the users’ ire at times.
The next set of Chatbots (second generation) uses an extensive word-classification process, which is a machine learning (ML) technique that assigns a set of predefined categories to open-ended text. For example, complaints to a company’s website using Chatbot can be organized by urgency, sentiments and replies given from a database. For instance, such a Chatbot after reading “Laptop screen was damaged when the parcel was opened”, will generate tags like ‘Laptop’, ‘damaged screen’ and will respond with a message of regret (understanding the sentiment) while internally pulling details of the laptop purchased by the buyer from the database. The Chatbot mimics a human agent and can advise replacement, schedule pickups, etc., like a human agent.
The next set of Chatbots (third-generation) uses natural language processors and sophisticated Artificial Intelligence (AI) to understand what is said or typed. Their conversation is more human-like, and such Chatbots keep on learning from past conversations. This brings us to the next question whether Alexa or Siri or Cortana or Google are Chatbots or not. Though the interface is different, they work on the same technology and thus may be termed as Chatbot.
Why is Chatbot important? Chatbot applications have successfully streamlined interactions between people and services, enhancing customer experience. They offer organizations new opportunities to improve the customers’ engagement process and operational efficiency by reducing the cost of customer service. As per a study conducted in 2020, 1.4 billion people use Chatbot, which can answer 80% of standard questions. 67% of customers used Chatbot last year. People around the world are conversing with Chatbots. However, the USA, India, Germany, UK and Brazil are the Top 5 users. Another benefit quoted by business leaders is substantial improvement in customer satisfaction (CSAT) scores.
While Chatbot came initially in English, many avatars in Indian languages have developed. A Bangalore based company has developed a Chatbot that can support 10 Indian languages. During the pandemic, hospitals faced a huge influx of inquiries about the coronavirus, symptoms, hospital beds etc. Hospital staff found it almost impossible to deal with every query in time. An Indian company launched a Chatbot that answered covid-19 queries and conversed with people in nine Indian languages.
Government agencies have not been far behind in using Chatbot. MyGov started Corona Helpdesk, which is a Chatbot that works through WhatsApp. GST portal’s GITA (GST Interactive Technical Assistant) provides answers to a large number of queries on GST. NIC has developed Chatbot named VANI, which is being used by a few ministries. However, much more needs to be done to make full use of this technology to make available an option to average Indian to speak to a Chatbot in their mother tongue and get information from Government websites or find the status of their application or to lodge a grievance.
Source: Economic Times