Let Chatbots do the talking


The Chatbot go by different names in the market. Talkbot, Bot, chatterbot etc. are some of the popular names.They could be general conversation based or they could be goal oriented. This article is based on the inputs provided by Himani Garg, Software engineer, Intuit while explaining the importance of chatbots in the times to come.

Are Chatbots for real?

Himani Garg, Software Engineer, Intuit

Chatbots where originally designed as conversational programs which will try to impersonate the dialogs by a machine to a human. The human, on the other hand, will have an immersive experience of conversing with the system in real time with expected responses which create an “illusion of understanding”. Inherently they are coded by humans and some of them learn using the help of Artificial intelligence in general.

Yes, they are different

Training a chatbot could be divided into two main divisions based on how they function. One would be called the Rule-based chatbot and other would be AI and ML trained chatbot. A rule-based chatbot is coded to give the user very specific responses. It could be as good as the programmer codes it to be. If this chatbot receives a bit of conversation that it doesn’t match with its subroutine, it gives a programmed response and doesn’t try to make it anymore better.

On the other hand, we have chatbot powered by Artificial intelligence. In simple words, they understand the natural language, convert it into machine understandable one and give the best response and try to make it more human-like.  They have the capability of learning from conversations and in the process give you smarter responses. One can easily feel the differences while “chatting” to this chatbot. 

The basics explained

Chatbots must have a sound dialog system coded into it to make as interactive to humans. The input sentences are parsed, and each chunk of the messages are broken into separate message objects.  In almost all the chatbot architecture there is something called normalizer that normalizes the message object and even tries to expand certain abbreviations supplied to it. The algorithm will decide the possible output and based on the priority the response will be given back. The intent of the chatbot needs to be developed by the chatbot and based on the priority. Now complex algorithms are built into place which mimics the natural learning of a child and learns based on the inputs. This is where it gets smarter in all perspectives. Chatbots that use neural conversation based dialog systems work on two basic strategies. They could be general conversation based or they could be goal oriented. General conversation based Artificial intelligence (AI) chatbot would talk to people on various topics that would put across in their conversations with the chatbots. The later which is a goal-oriented chatbot would help accomplish a real-world task like setting your alarm or even booking a cab.

Not simple to code

The most popular online voice assistants available in the market make it look very simple, but they are not. They are products of decades of research and a lot of linguistic computation going into the system. Various trials and errors, natural language learning and processing, context-based responses take years of training the machine. Chatbots are not just confined to giving you customer care assistance on a website or on an app. There are newsbots that can deliver news and updates, there are scheduling bots that help to manage meetings and to send info to teams involved. There are bot services that can give u financial advice and probably life bits of advice so that you live a better life.

Are chatbots future ready?

The Chatbot would be the best communication interface as the world is shifting to a more personalized message-based communication.  Organizations are adopting it rapidly as has proven to benefit every customer query instantly proving good for business too. The product service experience goes up manifold while saving money and time. Today there are chatbots that also evaluate vision and gesture movements.

While chatbots are being applied on new use cases, human intervention is still unavoidable.Server support and infrastructure issues could be looked at some key areas where they can assist on an emergency basis. Incident management and its automation are in experimentation, but they would prove very useful going forward in bigger corporations.

For more such interesting stories, read more.


Please enter your comment!
Please enter your name here