Chatbots: An Introduction to Conversational UI
“Chatbot” is currently one of the most common buzzwords in the industry. Many of us roll our eyes and think of it as another gimmicky piece of tech that has no real practical application. The truth is, when built properly, they can offer real value and we are going to see a lot more of them. We may even start to feel that they have become an indispensable part of our lives. In this article, I will discuss what a conversational user interface is and how they apply to chatbots.
What is a Conversational User Interface?
A conversational user interface (CUI) is an interface that allows users to interact with humans or bots using language, whether it be text or speech.
Humans and bots interact using conversational UI. (2018). Retrieved May 16, 2018, from https://dialogflow.com/docs/getting-started/basics
What is a Chatbot?
A chatbot is a program (typically using Natural Language Processing) that uses conversational UI as its mode of interaction between the user and a service. Chatbots and virtual assistants are sometimes referred to as “conversational agents” because they are the brains and design behind the interface. Bots interact with humans through conversational UI using language.
The most common types of chatbots are messenger bots, web chatbots, and virtual assistants which are nudging their way into our lives day by day. Traditionally, their use has been limited to customer support but the technology has matured and businesses are finding new opportunities in sales and marketing.
So why would anyone want to talk to one of these things in the first place? Our traditional experience with chatbots and virtual assistants is that they are procedural, scripted and lack the context necessary to understand the intent behind what we are saying.
Retrieved from https://siderite.blogspot.com
The problem with chatbots is that users tend to get frustrated and give up because the bot doesn’t understand what they want. Sometimes, users have to adjust how they naturally speak to fit the conversation design of the bot. Some of this is due to poor design and some of it is due to the limitations of the technology being used.
NLP to the Rescue
At a high level, platforms like DialogFlow and Wit.ai bring Natural Language Understanding, a subset of Natural Language Processing (NLP) that focuses on reading comprehension and semantic analysis. NLU helps programmed systems understand the context and intent of our language input.
Retrieved May 16, 2018, from https://cloud.google.com/natural-language
An example of when a chatbot would need to understand intent and context to take the correct action is a statement such as, "Wake me up in two hours." The system can infer that an alarm should be set two hours from the current time (ex. 9:15) without those values being explicitly stated.
This is a huge step. Systems are no longer just matching text input and procedurally returning a response, but instead grasping the context of our words to derive meaning and take action.
Advantages of Conversational UI
Assuming we can communicate with these bots in a way that seems passably natural, why would we use them over other modes of interaction? The primary advantage of conversational UI is its capacity to leverage the inherent efficiency of language. Below are just some of the benefits of conversational UI.
Easy to learn how to use
People already know how to talk, so the interface is intuitive. A user does not need to learn how to use a CUI.
Able to understand context and infer meaning
When we’re speaking, we can infer a great deal of information based on the context of the conversation. Contextual clues help us to differentiate between phrases that are vague or that have multiple meanings. For example, one might say, “Turn on the kitchen light” followed by “turn it off”. The phrase “turn it off” could refer to any device in our household. It is in the context of the kitchen light that we know what to turn off. One could then say “turn on the TV” and then “turn it off” and the same phrase would express a different meaning because of the new context.
What’s great about conversational interfaces is that they're able to understand meaning based on context (with the help of NLU), freeing us from explicitly stating every detail. So a chatbot can infer the intended meaning that we would like the kitchen light turned off, just as you or I would.
Allows for personalized experiences
Conversations can take many directions depending on a user’s input and preferences. Conversations often start off generic but as they progress they quickly become unique experiences. For example, a simple question like “How’s business?” could get completely different answers depending on your role in the company (Johnson, S. Chatbots, Conversational Interfaces, and the Rise of Messaging Platforms). The benefit of CUI is that the same interface can present information in completely different ways. User responses, preferences and CRM data can all be used to present a more segmented, personalized experience.
Human language is not procedural
The idea of “deep linking” or “shortcuts”, where a user can supply all the information necessary, skipping all the sequential steps that a Graphical User Interface might make one go through. Think of a form that sequentially asks questions, one after another. When using language users can combine all the required values in one step.
E.g. “I would like to reserve a cargo van for 1 pm this Friday”. This statement essentially combines a bunch of procedural steps together.
Q: “What would you like to do?”, A: “Make a reservation”
Q: “What would you like to reserve?”, A: “A cargo van”
Q: “What day would you like to reserve it for?“, A: “Friday”
Q: “At what time?“, A: “1 pm”
Conversational UI saves us time by allowing us to provide all the required information in one go.
Advantages of CUI with Speech (VUI)
The advantages become even more apparent when looking at Conversational UI with speech or Voice User Interfaces and virtual assistants. For one, we are more accustomed to VUI not being human (“Computer: Earl Grey, Hot”) so we are a little more forgiving here. Where VUIs really excel are in hands-free and multitasking situations, where the use of our hands is less optimal or simply not possible. Examples would be when driving a car, watching TV or cooking.
Retrieved May 16, 2018, from http://www.startrek.com/database_article/earl-grey-tea
In fact, using speech takes a third of the time it takes to type something and navigating through a GUI adds even more overhead to the number of actions needed to perform a given task. This becomes clear with the dangers involved when using a cell phone while driving, as opposed to the much safer option of using hands-free controls. One can multitask, without significant context switching, reducing risk and maintaining focus on their primary task. This overhead (pick up the phone, navigate to the app, navigate to the menu item, navigate to action item etc.) becomes even more burdensome when dealing with micro tasks like setting an alarm or reminder.
Write Once and Deploy Everywhere with Multi-Channel Chatbots
The reason it’s important to understand the abstraction of conversational UI is that any platform that supports CUI can be integrated with the same chatbot. For example, the same chatbot can integrate with Google Assistant, Facebook Messenger, Kik, Slack, web client, etc. Furthermore, users can interact with the same service across multiple devices: TVs and set-top boxes; virtual assistants; phones and tablets; desktop computers; cars and even watches. This is an extremely cost-effective way for businesses to achieve a multi-channel presence.
Conversational UI provides a consistent interface across multiple platforms and devices. Users can have uniform multi-channel experiences and businesses can maximize their reach at minimal cost.
Messengers have been around since the inception of the internet, what has changed is that with the advent of modern Machine Learning we are now able to write programs that can converse with people in a meaningful way. This is a paradigm shift in how we interact with technology. New methodologies for product design, user experience and conversation design will need to be adopted. Even bigger challenges are cultural adoption, vendors understanding the nuance of the landscape and clients understanding the value it can bring to them and their customers.