Chatbots: Where’s the Design-Centric Thinking?

essidsolutions

Think about the last time you had a delightful experience with a chatbot. Maybe it was something the bot did or said that made your work easier. Or perhaps it did something that made you go, “Wow!” Have you ever come across such a chatbot? 

Your answer is likely “No.” Yet, countless companies and research labs are toiling away at building chatbots. Most consumer-facing companies today have some variant of a chatbot either on their website, messaging platform, mobile app or some other channel. Chatbots are generally good at scheduling appointments, booking tickets, and processing orders or refunds, but few seem capable of delivering more than that. Fewer still create a memorable experience. 

See More: Say Goodbye to Chatbots: AI Powers the Next Generation of Conversational Assistants 

Not-So-Intelligent AI

The crux of any chatbot’s functionality is understanding the user’s intent and performing the right action. Today we have GPT 2 (and 3) taking the world by storm, showing state-of-the-art results on all NLP benchmarks and even fooling humans in text generation. It’s wrong then to assume that we aren’t there when it comes to chatbot tech, especially since intent recognition is a smaller task compared to what NLP can do today. NLP has been commoditized for the past two years and will continue on this path with folks like Hugging Face and Rasa democratizing AI. 

There are certainly examples of efficiently trained, human-like chat and voice bots delivering info and even empathy when needed. The potential for assisting humanity is limitless.

But why do we all unanimously dislike talking to chatbots then?  

Websites Are Graphical UIs

When a business creates a new website or a mobile app, a lot of time is spent on the design, look and feel. UI/UX designers and developers are hired to work out every minor detail, chalk out the wireframes and build the site out after a decent amount of usability testing. Why bother with all this? So users can have a good experience. So they can find what they’re looking for and simply get their job done. 

Websites and apps are merely a graphical interface for users to interact with a business. You might remember the websites in the 1990s ranging from tacky unusable ones to functional yet unappealing ones. Over time, the proper guidelines, tools and frameworks evolved not just to make websites and apps more usable but also a pleasure to interact with. We realized that tech is merely one of the enablers and that an equally important pillar affecting user engagement is the graphical UX.  

Chatbots: A New Type of User Interface 

When users interact with a chatbot, they try to find something or get something done. In other words, a chatbot is a type of conversational UI similar to how a website is a type of graphical UI. Today, many chatbots are the equivalent of those websites in the ‘90s–clunky, tacky, somewhat functional.

Why is this? 

Because conversational UX has been largely ignored in the bot building process. Conversational UIs are an entirely new paradigm–a new, more natural way to think about human-machine interaction. We’ve been down this road before when we moved from command line interfaces (CLI) to graphical user interfaces (GUI), from keyboard and mouse to touch screen interfaces, and so on. Every time we came up with a new way to interact with machines, we had to rethink the concept of user interfaces. We had to devise ways to make the new experiences work. 

The goal behind creating new interfaces is to make human machine interaction even more intuitive. With CUI, the idea is that conversations are an intrinsically intuitive way to do that, because conversing is something everyone knows how to do. However, as interfaces became more intuitive for the user, designers were challenged to come up with even better experiences. More thought had to be put into how icons should appear as we moved from CLI to GUI. For example, how can zooming be more intuitive on a smartphone? Pinch to zoom! 

See More: 5 Ways Chatbots Can Improve Workflows in the ‘New Normal’

The Conversational UX Is Broken

If conversational UIs are a new form of interaction, why are we not spending the same amount of time and effort rethinking the UX? If anything, even more time and effort should be put into perfecting the UX because users automatically expect intelligent bots to serve them in intelligent ways. 

So why do chatbots completely forget my previous interaction with them after just a few minutes? After I place an order and visit the website the next day, why can’t the bot proactively pop up and tell me whether my order has been shipped or delayed? Why should I have to log into yet another account to find out? 

If we know from websites and the millions of web ads that run today that copy matters–a lot–why are we not spending more time testing out different flows and conversation paths for bots? Why must chatbots go into the dreaded “can you please enter a valid input“ loop, asking the same thing repeatedly?

Today’s chatbots are merely glorified web forms because we’ve never bothered to reimagine what it means to interact with a conversational UI. Simply throwing a deep learning model at this problem will not make these issues go away. They require careful thought, deep-rooted empathy for the end-user and design-centric thinking to craft elegant conversational experiences carefully.   

The Chatbot Platform We Need

Conversational UX is the missing pillar. Today’s chatbot platforms throw many tools at the bot builder with the expectation that the builder will somehow figure out how to create a solid conversational experience for users. This is the ‘90s equivalent of providing the domain and infra and asking a website builder to write the website copy, do the graphic design, learn to code, learn UX and then build the website. Sure, some folks painstakingly did it, but in the end, all this left us with was a bunch of bad websites.

 Providing tools is not enough. Making sure the tools are used in the right way is key. Who is to be held accountable for the plethora of unusable bots? Should we pin the blame on the bot builder? No! While bot builders need control over the user experience, asking them to design it from the ground up is an unnecessary burden. Who wants to worry about that? No, the blame rests on the platform itself. It’s the duty of every platform to ensure that every bot built on top of the platform is one that by default provides a great conversational user experience. If you’re providing the tool, why not provide it with guardrails and guidelines? No bot builder wants to build a terrible bot, but if it takes a ridiculous amount of time to build a good one, they’ll have no choice.

 We live in very exciting times. Never have we been at the intersection of so many converging technologies, all moving towards making our lives easier. Conversational UI is without doubt at the forefront of this wave. However, it’s clear that we need to start looking at chatbots as a whole new category of UIs by keeping the user at its center. 

What are you looking for in a chatbot? Have you had a meaningful exchange with one yet? Share with us on LinkedInOpens a new window , TwitterOpens a new window , or FacebookOpens a new window . We’d love to know!

MORE ON ARTIFICIAL INTELLIGENCE