Want your Chatbot to seem more Human? Use Intelligent Listening

Creating Chatbot Experiences that feel Human

Let’s face it. We’re terrible at texting.

One of the most significant challenges facing chatbots is how to make the conversation feel like a natural conversation we have on a regular basis via mediums such as Facebook Messenger or standard text messaging. Unfortunately, we humans have developed certain tendencies in the way we interact digitally that makes accomplishing this feat harder than you would imagine.

Look through some of your recent conversations via text message and Facebook Messenger, and you’ll quickly notice a glaring trend. We don’t always finish complete thoughts or responses in a single message. It is much more common to see one thought fragmented into a few smaller messages that together make up a complete response.

Trending AI Articles:

1. Intro to Deep Learning in Social Networks

2. Ultimate Guide to Leveraging NLP & NLU for you Chatbot

3. Deep Learning in 7 Lines of Code

Why does this matter?

If you chatbot interprets every single message from a user as a complete response, then it is going to try to respond the best it can immediately. Not only can this cause the chatbot to answer incorrectly by not knowing all of the necessary information, but it will break up the natural flow of the conversation and likely frustrate the user. Users will be more apt to use your chatbot if they don’t have to rethink the way they communicate.

This chatbot is a terrible listener, and the user is not happy about it.

Typing indicators to the rescue.

So how do you overcome this hurdle? Typing indicators of course. Yes, the same typing indicators that are synonymous with iMessages and chats in virtually every modern app in existence. For this to be successful, there are several important factors to consider.

1) Your chatbot has to be able to recognize whether or not the user is typing. If that capability isn’t possible, then intelligent listening is not going to work.

2) Set up your chatbot so it waits X amount of time (3 seconds is a good starting point) after a user has stopped typing before evaluating the recently sent messages as a complete response. While the chatbot is waiting display typing indicators either immediately or after a second or so to let the user know that the chatbot is still alive and working.

3) After the pre-defined wait has passed, the chatbot should assess the user’s recently sent messages as a complete response and determine how to best answer as usual.

This chatbot listens like a human normally would and provides a far superior user experience.

Give this intelligent listening feature a shot in your chatbot and see firsthand just how big of an impact it can have on the overall user experience. After all, there are few things in life more painful than interacting with a terrible listener. This holds true for your chatbot as well.