2016 is shaping up to be the year of messaging. 

We're only a few months into the year, and we've already heard from Facebook that its Messenger reached 800 million users, WhatsApp went over a billion and Kik, Line and others show continued growth. 

Many of these companies are pivoting to attract businesses to not just advertise, but engage consumers on their platforms. Facebook’s announcement of Businesses on Messenger for example, calls out a strategy for consumer support and transaction. Chris Messina, developer experience lead at Uber, sums this trend up well in his recent post, "2016 Will Be the Year of Conversational Commerce."

Which brings us to the rise of the robots. 

Messaging startups from around the world such as Mezi, Magic, GoButler, Operator and more are jumping on the bandwagon of the more established players with a seemingly new and simple idea: building “bots” that interact in everyday language. 

This new generation of human-machine interfaces (HMI) is posed to replace Graphical User Interfaces (GUIs) with something that has been around for a long time: text interfaces, which are still the standard for IT and computer programming professionals. Why? Because text has remained the most efficient type of HMI for them.   

Natural Language: A Look Back, The Road Ahead

Users no longer need to remember a multitude of cryptic short commands to accomplish tasks, get questions answered and receive information — now they can use everyday language. This opens up text interfaces for broad use beyond the IT professional applications. 

Text interface's goals have moved beyond simply opening, moving or editing files, to shopping for a pair of jeans, checking on an order status, paying utility bills, making appointments with the dentist, reserving a hotel room or ordering food, all through intelligent, automated text interaction.  

Starting in 1997, I spent five years studying computational linguistics and phonetics. These five years only gave me a glimpse of the subject's complexity. Language is one of the most complex systems evolution has ever come up with, and teaching computers how to speak and understand natural language is no small task.

Let me give you an example: The other day I tried out a virtual assistant currently in beta mode built into a retailer's website. The assistant engaged me in a conversation about jackets. 

It began with the prompt, "Where and when will you be using this jacket?" I typed, "I need one for a skiing trip to Mass." It then went on to ask if this was for a man or a woman. Here is how the system interpreted my responses:  

  • Gender: "male" 
  • Designed for: “skiing, cold, winter" 
  • Travel destination: “one” 

The assistant misunderstood "one" as the destination of my trip, versus the lovely state of Massachusetts. Yes, it is in beta, and you might say that I didn’t answer the question, since it asked specifically “where and when will you be using this jacket.” 

People Won’t Respond How You Expect (and Want) Them To

My interaction illustrates the challenges of human language: It’s highly ambiguous. 

Unfortunately, the engineers who have designed many of the text-based bots in the past, have studied the world of programming languages — which are the direct opposite of natural languages: no ambiguity of meaning, clear and easy syntax. 

We've seen this before in the world of Interactive Voice Response (IVR). Too many applications were built by engineers, who know very little about how language works as a means to communicate between human beings, let alone how to make machines use it.  

Learning Opportunities

Here's a classical example of an IVR system giving stock quotes: 

IVR giving stock quotes

“Microsoft” is a perfectly acceptable answer to this question, albeit not what the engineer had in mind when he or she programmed the system for a “yes” or “no” answer.  

Is There a Linguist in the House?

Flock of developers are jumping on this fascinating new thing called Conversational UIs, only to build bots that are mediocre at best, or counterproductive at worst. My concern is that we will quickly ruin this emerging field and make early users turn away and dump this new form of UI, telling others: "doesn't work, stick to the app.” 

We need help. 

I know you're out there: General linguists, computational linguists, psychologists, UI designers, human-machine interface specialists, actors, writers, authors. 

We need you. 

The industry emerging around Conversational UI needs you. You understand the complexity of language and communication and won’t repeat the mistakes of the past. And in doing so, you will create messaging experiences that truly help us with our stressful lives and build solutions that make things easier, not harder. Faster, not slower. 

Or as Steve Jobs said: “technology alone is not enough — it’s technology married with liberal arts, married with the humanities, that yields us the results that make our heart sing.”

Automated messaging solutions have the potential to disrupt how we work, do business, live our lives. Let’s get it right from the start. Let’s involve the people that can help us.

fa-solid fa-hand-paper Learn how you can join our contributor community.