Search
  • Kiera Newton

It Takes More than Just AI to Make a Good Chatbot

When it comes to chatbots, computers designed to have a conversation with humans, the design tends to be focused on AI. This makes sense - to a degree. Artificial intelligence is that great sounding thing that brings us one step closer to turning science fiction into science fact. Yet, chatbots require more than just AI in order to function properly. Their base need is something far simpler – the available information. A chatbot can only interpret information. If someone has a bad experience with using a chatbot, then the first step is to look at the information the robot can access. A chatbot – along with the AI powering it – are nothing without proper information.





Making Information the Heart and Soul of a Chatbot

Imagine that you’re at a party. You know that you need to make conversation and mingle. What you don’t know is what you could possibly talk about. You don’t know anything that would interest the other person. What you hoped would be a good time becomes an awkward conversation that becomes an uncomfortable silence.


This is what happens to a chatbot when they aren’t responding properly or asking the right questions of users. Chatbots don’t have the information they need to have a good conversation with the user. The fault isn’t with how the chatbot knows how to have a conversation, which is guided by AI. The fault was in the training of the chatbot and the information used to do so. This is why training with the relevant information has to be at the heart of the product when making a chatbot.






Information Begets Intelligence

Get started on the right foot by building the chatbot processes on two tiers. Things start out with the chatbot initiating the conversation, being asked a question, and then attempting to interpret it correctly and find an answer in it’s knowledge base. It then provides the answer back to the user. This is where the all-important second-tier comes in, and it’s where good chatbots are separated from the bad ones. If a chatbot isn’t confident that it’s able to come up with the correct answer for the question, then it needs to make a not and ask someone else (another human) for the answer. The human can then train the chatbot with the new answer. The chatbot then learns from the interaction and it has new knowledge that it can use to answer questions in the future.


This ultimately results in the chatbot providing a better service and overall user experience, and – much like humans learning – it’s a process that should never end.


Improving User Experience Improves Bottom Line

After getting to grips with the basics and getting the chatbot going, you can dispense them across many digital channels – such as your tenant portal, Facebook Mesenger, WhatsApp or SMS. This is where chatbots start to add extra value to tenants who need to get answers to their questions but don’t want to (or can’t) use the telephone or want to wait on hold for your overworked contact centre team.


Chatbots are able to answer queries, find an appropriate service for a user, and direct calls for them. No matter what they do though, the experience is the most important part. Chatbots are the wave of the future, but the only people who really win with them will be those who make it easy for their chatbots to access relevant information and learn new things.


Connecting Chatbots to Your Brand

The way to think about chatbots is as a way to for tenants to interact with their information and get support from their landlord. If the information that you give your chatbot is lacking, then so will the answers – and experience – they provide users. This isn’t good at all for connecting a chatbot to a brand. Once again, the key is with the information the chatbot has. All the artificial intelligence in the world means nothing if your chatbot doesn’t have any actual information to work with.

36 views

© 2019 by Fuzzlab Ltd.