Over the last few years, advances in Machine Learning and AI, have allowed Conversational UI to become more interactive and easy to use. Chatbots and Virtual Assistants rely on CUI, while more and more systems begin to take this UX model into consideration.
8. MOBILE APPS AVAILABLE IN 2017
AVERAGE COST OF MOBILE
APPLICATION DEV
MONEY SPENT ON GUI BASED APPLICATIONS
9.
10. Human Voice
Synthesis
Text & / or
Speech
Engine
Natural
Language
Processing
(NLP)
Machine and
Deep
Learning (ML)
Better Voice
Recognition
Hardware
DNN
Contextual
Responses
More Data
Voice
Variations
15. INTENTS VS.
SCREENS
Easier to implement, less
limitations
LESS
ERRORS
System errors can be
easily handled
CONVERSATIONAL
More engaging
FASTER
vs. GUI
AVAILABILITY
More channels
VOICE CONTROL
Can be used where hands
are not optimal
ADVANTAGES
OVER GRAPHICAL
UI
25. Q&A - ANY QUESTIONS?
YOU CAN FIND ME AT:
www.meydman.info
www.linkedin.com/in/meydman
michael@abmecorp.info
Editor's Notes
A little about myself. I wear a few different heats and my career has been split between entrepreneurship and management consulting. I see myself as a thought leader and spend a lot of time learning about new trends in technology and user experience. Over the past few years, I have been focusing more on emerging technology, which includes artificial intelligence, augmented reality and how our interaction with technology is undergoing a major shift.
About myself
Entrepreneurship & Consulting
Thought leader in new technology trends
Focus on emerging technology
I would like to spend the next 20 or so minutes talking about a shift from traditional graphical user interfaces towards what is called conversational UI. I would like to put this is a large frame of UI evolution. As you can see from this somewhat exaggerated illustration, our methods of interaction have evolved from simple text based interfaces, to graphical visual representations, to touch, and ultimately towards voice based interfaces, which is what I will focus on.
Shift from GUI to CUI
Part of UI evolution
To give you a quick agenda, so you know what to expect. I will briefly cover some of the current limitations in graphical interfaces, then talk about recent advancements in machine learning and AI, which allowed conversational UI to become more widely used. I will then go over a few use cases of successful implementation of this type of interface, and will hopefully leave a few minutes in the end for Q & A.
Before we begin, how about we get a little bit of an idea what Conversational UI feels like?
Jeopardy
Weather
So let’s talk a little about limitations of current graphical interfaces. GUI works by presenting all available options on a screen in front of you. WYSIWYG. How many of you have worked on or at least saw some of these screens? How many payment methods do you need to shop online? What about inherently complex screens used by in music or video productions? Stock trading? The complexity here can be mind boggling.
GUI presents all possible options
Too many options to choose from
Complex screens in music & video production, stock trading
Mind bogglingg
So what are these challenges? #1 issue is system unpredictability. I think most of you know what a blue screen of death looks like. Sure, it can be blamed on bad QA, or coding, or even Project Management, but the bottom line is the user left without ability to interact. The complex screen mentioned in the previous slide, even as our screens have grown to almost stadium sized proportions, we are still struggling to put all of the functions into a seamless experience. Some, like Apple have figured out a few things, but the majority of enterprise user interfaces are still clunky and unwieldy.
Blue screen of death – unexpected behavior
Complex UI even though screen real estate has grown
Only a few figured it out, Apple
Let’s talk about the cost involved in designing and developing these interfaces. In 2017, there were over 6.5 billion mobile applications available on the market. The average cost to launch a mobile application is around $10K, which means the worlds has spent over $65 trillion dollars on building applications with graphical user interfaces. This sets the context on how much money is being spent on UI.
While all these resources have been used to fine tune GUI, the technology which drives conversational interfaces had leapfrogged in the last few years. It was able to do that with the help of data science.
While most resources have been focused on GUI
Data Science had leapfrogged over last years
To be more specific, improvements in Natural Language Processing or NLP have paved a way for more fluent and human-like interaction. Deep Neural Network is one of the models which have shown significant potential and is at the basis of Google’s Cloud Speech Engine, which has recently passed a test for human-like speech synthesis. Improvements in hardware-based speech processing and increased availability of training date have also contributed to drastically better voice recognition, including accent variations and contextual communications.
NLP became more fluent and human-like
Deep Neural Network -> Google Cloud Speech Engine -> flawless human speech
Hardware, accents & contextual communications
So what is Conversational UI? The simplest way to define it is an interface which allows a user to interact with a computer system using spoken or written language. Unlike more traditional Graphical UI, which uses predefined visual representations of what user may see or react to, CUI allows users to chat, picking up input directions from the context of a conversation.
Spoken & written communications
The first conversational interface was launched in 1966 in an effort to create an automated therapist. Unfortunately it proved too narrow focused and ahead of its times, when most people did not have access to computers. In 2010, Apple launched its Siri assistant as part of the iPhone platform. Although initial response was mixed, it gained traction and became an inseparable component of iOS. Wide availability and portability of Smartphones became the catalyst for this technology. Today, millions of people rely on Instant Messaging and Social Media tools to communicate, presenting a huge potential for conversational interfaces.
1966 – Eliza – automated therapist
2010 – iPhone Siri
2018 – Conversational UI
For clarity sake, Conversational UI is divided into two main communication modalities, which can be used independently or in conjunction with each other. Written conversations, such as the ones taking place in email, instant messaging tools and through social media channels. Spoken communications, primarily taking place in Interactive Voice Response systems, and more recently, through Virtual Assistants like Alexa and Google Home.
Let’s also understand the various channel where these modalities may be present.
There are a number of advantages CUI has over GUI. Unlike traditional screens, CUI uses a concept of Intents, which are much easier to implement than complex graphical screens. System errors can be handled with more ease, virtually eliminating system crashes. Being conversational is more engaging, it’s that simple. It is also much faster as we can see from these examples. Another benefit is availability through multiple channels, the same system can be accessed as a standalone application, be available through most instant messaging tools and also present as a skill on Amazon’s Alexa. Which brings me to another major advantage, voice. In situations where hands are needed for other tasks, like driving, the only viable option is through voice.
www.youtube.com/watch?v=3rjKLeG5cRU
www.youtube.com/watch?v=ZZsebw4pPmg
Quick video illustrating the difference between GUI and CUI in online shopping.
I would like to make one point clear. While I am a big proponent for conversational UI, it is not the ultimate solution. Until we learn how to interface with technology using our thoughts alone, the best interaction can only be achieved through combination of interfaces. The right mix of visual and conversational messages depends on the type of information being communicated and the expected response.
Input perceived in many ways
Combined interfaces mix visual & auditory
It’s not just about Chatbots, it’s a shift in how we interact with systems and technology. Both Alexa and Google Assistant provide a marketplace for third party applications. If you think about what Apple’s iTunes Store did for a third party app economy, the current opportunity exists for Conversational UI based applications. According to latest statistics the list of available skills on Alexa already grew to over 25K in just a year. I believe there is a huge opportunity there.
Third party marketplaces similar to App Store & iTunes
Over 25K skills and growing
Potential for opportunity
Now, I would like to go through a few interesting examples of how Conversational UI has been implemented and used in different contexts.
When Facebook opened up their Messenger app to developers in 2016, KLM Airlines were one of the first ones to realize the potential and launched a conversational customer service app through that channel. After a year in operation, they are reporting to have increased over 40% of mobile interaction, with 15% of boarding passes being used in this way, resulting in 5 point increase in Net Promoter Score value, which evaluates customer satisfaction.
https://www.facebook.com/business/success/klm-messenger
Microsoft launched this bot in China’s Weibo social platform. It instantly became wildly popular. In 3 months after it’s launch it had 0.5B conversations and gained over 850K followers. Xiaolce, which means Little Ice was even able to pass the Turing test for whole 10 minutes.
Virtual Assistants - capable of multiple tasks, support third party applications, and use probably the most advanced speech recognition technology available today. Their growing list of features includes almost every aspect of our daily routine.
Other examples of successful implementations of Conversational UI includes a large number of specialized chatbots. Most of these bots are able to engage in chit-chat, and understand a constantly growing list of request. For example a company called x.ai has create a bot called Amy which can completely replace someone’s receptionist and handle all meeting and scheduling requests. A bot called Operator can help you find pretty much anything on the internet, especially if it’s for sale. There are also News bots which can understand your news preferences and even a virtual psychologist who can help with anxiety and depression.
Here are some of the tools and resources to help someone start with Conversational UI.