Posted on: July 21, 2020 Posted by: admin Comments: 0

Artificial intelligence is changing the way we search and get things done and Chatbots are the real example of human aspiration to get rid of doing things which doesn’t excites them much.

Recently I have been learning a lot of about AI powered chatbots and got quite fascinated by amazing things it can do without involving humans. There are chatbots which can talk you like your girlfriend /boyfriend. There are weather chatbots like Poncho The Weather Cat which can tell what’s weather outside in flash of a second without requiring you to type.

The most popular use case of chatbots have been on the enterprise side customer support. It has literally evolved to the extent that 70–80 % conversation is being handled without involving real humans . BFSI sector are already leveraging chatbots heavily to address their customer queries and assisting them to quickly go through their banking transactions.

Home automation bots in collaboration with Alexa and google assistant can seamlessly help user to order foods, groceries, get all their stuff done. Travel, health, education, agriculture, e-Governance, every Industry Imagine possible are potential use case for these intelligent bots . They can help you sell more, engage your customer, answer your customer queries, educate your customers.

So what this chatbot is all about? Is what i would demystify today. We will be covering following key aspects in the next few minutes

1.What Is Chatbot?

2. How It Works?

3. What Is The Technology Behind Chatbots?

4. Types Of Chatbots

5. Facts & Figures For Chatbots

6. Challenges, limitations and opportunities

So are you ready to walk along with me .

So Let’s Get Started,

1. What Is Chatbot ?

I often quote:

Chatbots are the innovation done by intelligent humans to serve humanity intelligently. It’s the most advanced & simplest way humans can interact with computers to get their things done.

As per Wiki:

A chatbot (also known as a smartbot, conversational bot, chatterbot, interactive agent, conversational interface, Conversational AI, talkbot or artificial conversational entity) is a computer program or an artificial intelligence which conducts a conversation via auditory or textual methods.

Conversations, whether chat-, text- or voice-based are going to be more pervasive moving forward for task completion — Aakrit Vaish, Founder & CEO Of Haptik

Asking questions has been the natural way for us humans to interact & acquire about anything they wanted to understand.

A chatbot is kind of a computer program that fundamentally simulates human conversations. It allows a form of interaction between a human and a machine the communication, which happens via messages or voice command.

So with the internet and search engine revolution, we humans started to find answers for everything imagine possible on Google, but it was not solving all our curiosity to get things done. Google only managed to give you information but not lead to end outcome. This urge to get more things done quickly triggered the development of Chatbot .

Let’s see how chatbots evolved. coined a very informative infographics below which will help you understand that Chatbot has been existent in the past, but recently with the right market condition to help it thrive it has become a norm for many B2B businesses.

Chatbot Evolution:

Chatbots are the new world of human to computer interaction where humans and AI bots are helping each other out .

How Chatbot Works? What is the technology empowering Them ?

Most of the chatbots are a kind of messaging interface where instead of humans answering to your messages bots are responding. fundamentally they may look like another app. But the catch is that UI layer where you interact. The conversation humans have with bots is powered by ML algorithms which breaks down your messages into human understandable natural languages using NLP techniques and responds to your queries similar to what you can expect from any human on the other side .

For Example :

(I would present one small use case form Haptik app which is one of the largest AI based conversational chatbot app from India. )

You can get the app here :Haptik Assistant – Reminders, Flights, Daily Quiz – Apps on Google Play
“Haptik is the one app every Indian must have on their phone” – The Times of India, NDTV, Economic Times, Business…

Suppose you wanted to travel from Delhi to Indore. You can simply open Haptik Chatbot app and type a message

“ Book a flight from Delhi To Indore. “

You will be getting an instant reply asking

“ for how many people you want to book a flight ”

Screenshots from haptik app.

And once you respond, bot will respond back with all possible flight booking details in a flash Isn’t amazing. As you can see, the response this bot gives back looks quite natural, the way you would have expected from customer care executive.

But hold on, there is lot of heavy-lifting works which goes behind, to give you this kind of experience.

The very first thing a bot needs to do is to, understand what humans have given as an input. This understanding is achieved through various parameters such as as Text Classifiers, Suitable Algorithms, Artificial Neural Networks and Natural Language Understanding (NLU) it’s a type of NLP. All these factors contribute to the overall bot functionality and intelligence of the Chatbot. Once you have understood the intent you need to respond back with appropriate message which should sound like a natural reply. To achieve this, another type of NLP called Natural Language Generation(NLG) technique is used.

Let me take you all through a high level understanding in the next sections,

Text Classifiers:

Here we primarily segregate words and sentences into a meaningful intent. This intent is understood by the Chatbot and accordingly helps it to respond.

For instance:

If you type how are your doing ? the associated intent would be “ i am doing great” or “I am good ”. The Text Classifiers enables the Chatbot to classify information and thereby produce responses based on the same. It is crucial that the Chatbot is able to distinguish the input cases and generate meaningful and effective responses.

Natural Language Processing :

Most of the research being done on natural language processing revolves around search, especially enterprise search. Bots are super dependent on this AI branch called NLP. Human language tends to go very chaotic and complex and NLP engine comes handy to tackle this mess. It is made up of a number of different libraries, the NLP engine does the work of identifying and extracting entities, which are relevant pieces of information provided by the user, using libraries for common NLP tasks like Tokenization and Named Entity Recognition(NER)


Given a character sequence and a defined document unit having sentences, tokenization breaks it up into pieces, called tokens , perhaps at the same time throwing away certain characters, such as punctuation.

For instance :

Input: Friends, Romans, Countrymen, lend me your ears; 
Output: This sentence will be breaker into 7 token words.

  1. Friends
  2. Romans
  3. Countrymen
  4. lend
  5. me
  6. your
  7. ears

A token is an instance of a sequence of characters in some particular document that are grouped together as a useful semantic unit for processing

NER: Named Entity Recognition:

While building any conversational bots/ Dialog system one can employ the following approaches to do so

  1. Generative Based
  2. Retrieval Based
  3. Heuristic Based

All of these approaches rely on NER some or the other way

NER is a subtask of information extraction that seeks to locate and classify named entities in text into predefined categories such as the name of a person, location, organization, contact detail, expressions of time, quantity, monetary value, percentage, etc.

Most of Entities being retrieved falls into following broader category

  1. Numeral: Detecting numbers
  2. Temporal: Detecting with time
  3. Pattern: Understanding patterns and regular expression like email, mobile number etc
  4. Textual: Detecting entities by looking at the dictionary

While NER looks for entities in pre-defined categories (for example, place names or addresses). They might also use a library called a Normalizer, which catches common spelling errors, expands contractions and abbreviations etc..

I have covered NLP in detail in one of my previous article. Please refer below to know more

ViceBots, Future And Type :

What we have discussed above has to mainly to do with Text Based conversational chatbots where messaging is the key medium to serve customers. But the future is going to be beyond text based user engagement. Most of the people are not comfortable typing and that audience is quite large. Typing is the limitation but everyone knows to talk at least in their local language. So future lies in making a Voice based conversational chatbot which can understand in multiple languages and respond back in the respective language which user is comfortable talking.

Having said that our system should be fault tolerant & reliable, as voice based conversation can go horribly wrong, user can’t navigate back and redo the entire conversation, so where you augment and mix text based approach has to be well thought through.

Success of Voice bots will be mainly be decided by how robust is it’s workflow design and how intelligently it can fall back to humans and text based conversation in case of any discrepancy in conversation.

Type Of Voice Bots :

Voice Bots can be bifurcated into two broader category

  1. Hybrid Bots : Text+Voice based
  2. Voice Only Bots : Voice controlled devices

Voicebots behaves like conversational agent who acts like a personal assistant taking up your calls, reading messages, alerting about important events etc. They also integrates well with potential voice interface. A voice-first device is an always-on, they are intelligent piece of hardware with primary interface for interaction being audial both for input & output.

Text Vs Voice

Even though Voice is the most natural way of conversation but text due to huge acceptance of text based messaging has lead the way for Chatbot industry. with more than 3 billion active user across top 4 messaging app the opportunity to convert them into paying customers has been the main focus area for most of the business leaders.

Although, Voice and text bots have different input mechanism, they both are powered by NLP engine to function intelligently. An additional layer which serves a voice bot converts user’s voice into text before forwarding it to NLP engine. Similarly, the response is also converted into speech as output.

Chatbots Facts & Figures:

Subro has compiled some interesting facts and forecast for AI based chatbots , which i found compelling and so sharing below.

What are the challenges for chatbots?

Having understood how chatbots are catching up popularity with its user friendly interface and its power to harness messaging as a platform to engage customers. it is very important to understand what are some of the core challenges this chatbot world is facing. So that we can measure the opportunity in the current context and beyond.

So let’s see what are some of the major challenges lying ahead for adopters of Chatbot technology.

1. Chatbot Security:

In this age of data privacy and sensitivity it is very imperative that user can trust on the bots with whom they are sharing their data. While engaging customer the design of the chatbots should ensure that only relevant data is being asked and captured as an input and also is being securely transmitted over the internet. That trust, that care that it is absolutely safe to share data with bot needs to be a part of the Business DNA who are going to adopt this Chatbot tools to serve their customer needs.

Companies needs to assure that no hacker gets an access to chat interface as it will lead to a total disaster for the company as many sensitive data of the customers can be misused and mishandled.

How secure is your Chatbot, will be the primary concern for any user who are going to make the decision of using your bot interface to do any kind of interaction.

2. Making Chatbot stick and likeable.

The biggest challenges for Chatbot is that it needs to be quick and effective in answering customer queries. The lacklustre response and unwanted lags may force user to move away from the Chatbot interface and possibly never to return back.

The biggest limitation of any chatbot will be grabbing user attention and retaining them till completion.

So no matter how robust and technically superior your chatbots may be, if it is not designed for WOW user experience, it may fail miserably.

3. Understanding user sentiments & emotions in case of VoiceBots:

VoiceBots can go extremely wrong it it fails to sense the the user emotion and can mess up badly. Identifying user problem from the voice, breaking them into meaningful intent and responding with appropriate voice reciprocating the right emotions is not an easy task, it requires the bot to be trained a lot with real human sounds so that the intelligence to respond in right tone and pitch is built.

4. To build Vernacular Chatbot Is a Daunting Task, But Is important For Your Business To Succeed :

Only about 7.5 percent of the world’s people call English their native language. So in that case just making chatbots in english is not going to solve the real problem of user who wants to interact in their native/local language. In-fact having your bot interacting in multiple languages will be a win win proposition as it will impact the customer deep inside and will make them feel heard.

Just imagine VoiceBots talking to your customer in the same language which they always speak in their day to day life, will it not make them feel special. But it will have multiple hurdles to achieve perfection.

Effective communication requires cultural awareness and attentiveness to nuance — and effort. There’s actual work involved in building a bot that communicates effectively in multiple languages!

Making your chatbot Polyglot will require you to involve more human translators instead of machine. It will give you more meaningful training to the bots to understand culture nuances.

Challenges in adopting multilingual approach

While you adopt multilingual approach these are some of the challenges which your bot need to address

Non-standard language:

Humans very often make use of some nonstandard words like slangs, they very often misspells and also make use of emoji while typing so your bot needs to be able to respond appropriately. Working with native speakers is incredibly helpful here. The more you work with local language speaker the more reliable your multilingual will become.

Attitude and context:

When your customer/user engage in any kind of conversations they expect them to understand the cultural context and also the underpinning attitude of the language. Once again, a capable translation team can help develop a bot that’s culturally literate and is emotionally sound.

Language Specialization:

If your Chatbot needs to address to any particular type market where it is supercritical for it to speak their language it is absolutely recommended to have a master linguistic who specialises in that particular language. I may be sound in french but i am Indian by birth and have solid command over Hindi so you can rely on me for Hindi but for french irrespective for how many years i am using it.

You can’t let your bot fail and ruin your brand identity by being mediocre

What’s Next ?

I would like to end this part of the article on this important note

“Yes you will be challenged when you want to build something great which can change the future. But if you are committed enough and have right set of brains working together to solve the clear problem statement you will succeed, in the process you will experience some roadblocks but that will be more like a learning curve which will help you achieve the perfection ”

We will further look into AI based chatbot market size, opportunities and how some of the top startups and businesses are shaping the future world with their smart AI chatbot based solutions.

Leave a Comment