Before Conversational UI was Born
Before conversational user interface or Conversational UI or CUI become a buzzword in 2017, the younger generation may not know this is how the computer screen looked like. Remembered Command Prompt?
This was a text based experience, and required knowledge of computers to compute.
The user interface we are familiar with now is called GUI or Graphical User Interface. It makes computing easier, convenient, friendly, visually appealing and does not require extensive knowledge of computers.
GUI made computing popular, led to the invention of handheld devices making computing on the go possible.
However GUI has its own set of limitations in the age of internet when there is a rapid transition from desktops and laptops to mobile devices. A webpage to book your seats for the upcoming blockbuster movie looks different on mobile devices. Add to that the absence of uniformity – some essential features present on desktop version might be missing on the mobile platform.
Armed with an Yellow Page and a landline this task can be completed easier. But aren’t we more comfortable to deal with things all by ourselves without the need of human intervention?
What is Conversational User Interface
Any UI that mimics chatting with a real human is what conversational UI all about. In simpler words, replacing the need for GUI – like buttons, links, page changes, swipes and taps with simpler model of “get what you ask for minus the hassle of GUI”.
Why Conversational UI is the Future
Firstly, Conversational user interface is important because of the rising popularity of native chat users. Chatting platforms like Messenger and WhatsApp engage users like never before.
In a study it was found that 75% of the time is spent on such platforms making CUI an important aspect to consider and improve.
Secondly, a typical GUI requires moving from one screen to another, understanding the process and flow which can be quite tedious.
With increasing number of apps, an average user would need to familiarize and remember dozens of user interface as each of them are differently designed.
A well designed conversational UI can increase conversions by up to 2X in comparison to their GUI counterparts.
Thirdly, CUI has the ability to humanize the experience through the use of emojis, friendly messages, happiness among other things.
Giving a human touch on a computing platform would differentiate a good CUI from another.
Fourthly, it provides a quick communication medium by talking or typing out what we need rather than firing up the browser, searching, tapping and swiping. In today’s world of instant gratification, every minute matters and conversational user interface offers unmatchable speed.
Fifthly, it offers specificity of need. For example, if you have to book a cab, you would use an app to first tap on “get a ride” section. Next you would need to mention the destination and the number of occupancy. With different “book a ride” apps, you would find different user interfaces at different parts of the screen.
CUI makes it simple – you would simply mention in text
Book an ride for 2 from Indira Nagar to Madhapur at 2 PM
Conversational User Interface to the Rescue
Let us now consider an alternative, how great would be to type –
Reservation for 4 people at 8 PM at Barbecue Nation
And your reservation is done.
This is what conversational user interface or conversational UI is all about. It makes conversation with a computer easier in order to complete a given task without the hassle of tapping, swiping on a poorly designed interface.
Google Allo has started on a very positive note – I could drive to dinner to a Dominos using the CUI of Google Allo.
Heard of chatting platform WeChat? In China, most apps are integrated with this popular chatting platform. So while in North America people are downloading dozens of apps, in China one app is enough.
Expect messaging platforms to take advantage of a wide range of media – like media players, games, in-message payments, photography, and so on.
The future holds as “you chatting with the messaging app” rather than the current “you chatting at the messaging app”.
With an app that is powered with CUI, you can book a table at a restaurant, book hotels and flights, order a ride or even make bank payments.
Principles of Designing a Complete Chat Bot that Uses Conversational UI
The basic building blocks of a chat bot that implements CUI includes the following:
1. Designing the verbal communication
2. Designing the non-verbal communication
You can read in details about these principles on Medium’s process of designing a website chat bot.
The traditional design patterns of GUI does not work with CUI. Because in a CUI what matters is what you say and how you say it. The barrier between design and content disappears completely in CUI.
Matty Mariansky, creator of Meekan, a robotic scheduling assistant says
It’s the robot’s job to seize every opportunity to suggest the next step and highlight less-familiar features.
Being specific while asking a question is important. Users should not be asked open ended question.
So if an an e-commerce bot needs to know the size of T-shirt, it should ask and be specific like – “What is your T-Shirt size. We have small, medium, large”.
This way the user is spared from wondering if he or she should reply the size with cm or inch or small/medium/large.
Another case to consider might be beginning the conversation from where it ended the previous time. Making the conversation feel natural is an important element of conversational UI.
Another significant difference between GUI and CUI is discoverability. If you use an app today for booking flight tickets, you have an idea that you can book tickets. What about the other features like saving your favorite destinations, your e-wallet, passenger list, etc. You do not have an idea of these unless the developer has put them somewhere on the home screen or in your dashboard.
With CUI you explore as you go. For example, chat bots remember your travel list and favorite travel destinations. You would not need to hit a click or tap on the screen. If you get what I am saying.
In a Conversational UI, the bot has to seize every opportunity available to tell the user that it is capable of doing so and so tasks. From welcoming a new user to suggesting a task, to understand what the user is looking for and providing alternatives, the task of a bot is to make the user feel comfortable.
How Companies Can Strategize CUI
CUI require a clearly thought out strategy which includes the following:
Identifying the Service Offerings and Strategizing the Conversation
The first step is to recognize the utility.
Why would an user interact with your brand?
What are the most probable questions that the user can ask?
What response do the users expect in return?What service offering does your brand offers?
Transforming the service offerings into a conversation is the critical initial step to developing a conversational UI.
Choosing a technology that is quick to go live, easy to evolve is important for the success of CUI.
There should be a service layer architecture for core brand functionality. Select the right set of natural language.
It should be remembered that as CUI evolves and best practices emerge, the platform would need to go for significant changes to make it popular.
The chosen technology should not require massive changes to the organization’s backend systems.
Organizations need to form and implement a data strategy. As more and more customers use the platform, a single view of the customer should be considered to make learning more effective for the AI algorithm powering the CUI.
Organizations would need to think about conversational experience through text, voice, conversational flows, use of language, etc.
CUI requires rethinking of a company’s operational model. While servicing the customer where do intelligent agents stop and human intervention becomes necessary?
Does your organization have the right blend of CUI developers, data scientists, consultants who can make AI algorithms effective and more efficient?
Conversational UI and Artificial Intelligence
In 1950, Alan Turing designed a test which came to be known after him as Turing Test. It consisted of 3 players – A (Computer), B (Human) and C (Human).
The test included a series of questions and answers through which Player C would determine which of the other two players was human. Turing believed that if the computer could deceive the human (Player C) 70% of the time, the computer would be judged as more intelligent.
Over the last 6 decades the question still remains – can computers convince humans that humans are interacting with a human rather than a machine.
It has now been concluded through several tests that computers can converse with humans, but without full knowledge of what the conversation is all about.
Current AI has advanced algorithms to engage in a conversation without understanding much of the conversation it is having.
Humanizing Technology – An Example
Let us consider Amy – a chat bot application.
Amy’s job is simple enough. It helps schedule your daily meetings. Amy has access to your calendar. So when you have to schedule a meeting, you simply mail Amy, very much you would do to a human office assistant.
Amy has access to your calendar so she knows when you are available. She knows your favorite meeting locations and you can always send her a quick email whenever you discover a new meeting place.
However, Amy’s human nature comes to picture when she takes care of all your related correspondence.
For example, Arpita emails me
Hi Samrat, it was great to have a meeting on the latest technology trends for 2017. Do you have time to meet up tomorrow, or the day after or later this week? I am free mostly after 3 PM.
I can CC Amy and reply to Arpita:
Hi Arpita, I am available later this week. Please work with Amy (CC’d) to get this scheduled on the calendar. Thanks, Samrat
If you have noticed Amy has the ability to make conclusions that only humans could make earlier.
This is being done through deep learning. Deep learning trains software to recognize complex patterns in sounds, images and other data. It includes feeding large amount of data as input and refine response on the basis of a specific input.
Take a look at Amazon’s Watson
Conversational User Interface Apps and Websites
Some of the apps and websites that implement CUI include –
There are exciting things happening with organizations implementing conversational UI in their products and service offerings.
Don’t forget to check out this cute, sweet and entertaining robotic assistant called Peeqo
Skepticism around Conversational User Interface
Every emerging technology has its own share of skepticism. Many experts have highlighted the slow reaction times of the bots as a point of concern.
Many assume that CUI is designed to complete complex tasks. However some of the best bots designed in 2016 are those that are using smart chat design elements and algorithms leading to engaging conversations to quickly complete simple tasks.
Conversational UI Shortcomings and Future
We have already discussed in the previous section that bots can’t solve complex problems as of now.
Secondly, current chat botters don’t understand the sensitivity of the subject matter. Once, Siri almost led a potential suicide victim to a gun shop. However, when I tried the same Google Assistant Allo, it gave out a very welcoming and encouraging reply with a helpline number.
Thirdly, conversational UI is still following a text based communication model. Through text, bots do not have an accurate understanding of the human sentiments and emotions.
Future of Conversational User Interface
In 2017 and in the following years, expect CUI to come of age, get more matured with improvement in AI algorithms and best practices and heavier implementation of deep learning algorithms.
Primary focus would be on understanding text based sentiments through emoticons and app stickers.
Secondly, expect more research on interpreting emotion in voice. Can the chat bot understand whether you are happy or sad, disappointed or curious by your voice command?
Thirdly, leveraging conversational UI to interpret facial expressions.
Leverage conversational UI to study your body temperature – integration of chat apps with smart bands! For example, the smart band will monitor your heart beats, body temperature. Once it finds a significant change, it can trigger your chat app to message your doctor that you need medical attention.
At the very least, the chat app can inform you that you need medical attention and direct you to specialists near you. That is the future.
Research and implementation of emotional data is already on as you read this.
Conversational UI has just started rolling into motion and is still in its infancy. Few years ago, we witnessed transition from websites to apps. Year 2016 made a larger paradigm shift when it comes to consumer behavior as we have initiated a transition from apps to conversational user interface.
It will certainly take a couple of more years to fully understand best practices around conversational UI; the good news is that organizations like Facebook, Microsoft are investing more in bots.
In India, research and development of chat bots are at par with their US counterparts. Expect 2017 to be a watershed year for chat bots.