How to make conversation (with robots)

The Road to the Conversational Interface

The year is 2001 and Dr. David Bowman is running out of time. He’s trapped, floating in a metal capsule just outside of Jupiter, the only sentient being in light years. That is, if you don’t count the operating system that is trying to kill him.

“Open the pod bay doors, HAL. Hello…HAL, do you read me HAL?”

“I’m sorry Dave I’m afraid I can’t do that. I think you know what the problem is just as well as I do.”

The problem is that HAL 9000 has gotten a little too artificially intelligent. The supercomputer tasked with protecting human life now attempts to conceal an error, at the expense of its end users’ lives. The problem, as prophesied in Stanley Kubrick’s Space Odyssey, is that the machines of the future have become too smart, too intuitive… too human.

Things panned out a little differently off-screen.

In 2016, human-computer interfacing is not the chorus of disembodied voices Kubrick promised it would be. We click, we tap, we type. Wordlessly. Our machines, like the future-computers Kubrick envisioned, do a dazzling array of things. But why, when a mere telephone can bank, navigate, and find people mates, are we still mechanically typing into screens? Why, when computers can predict your biometric patterns, do we still push rote button combinations to get through to the DMV?

For all our advancement, seamless human-computer conversations remain in the realm of science fiction.

But as we’ll see, that won’t be the case for long.

Science fiction made real

As Kubrick’s 1968 blockbuster illustrates, conversational interfaces are not a new concept. In a time when Pong was just beginning to hit the stores, people were already dreaming of befriending (or being betrayed by) computers.

What Kubrick and society didn’t foresee were all the iterations it would take to get from punch cards to HAL 9000. That is to say, to get from interacting with computers like computers, to computers interacting with us, like humans.

Let’s briefly trace this trajectory.

From invention, computers have required humans to communicate with them on their terms. In the early days of computing, it was huge stacks of painstakingly coded cardstock. By the 70’s, command line interface had made communicating a little more intuitive, allowing programmers to use language (albeit highly codified) to control computers.

Within a decade, Microsoft had introduced Windows, reigning in the era of the mouse and the spatial, or graphic user interface (GUI). People could now communicate to computers with the click quick of an icon rather than long lines of code.

Computing continued to get easier. GUI’s became more beautiful. OS’s became touch-enabled. Now, 30 years later, we can tell computers what to do with the tap of a finger. The next step? Communicating without even lifting a finger.

To get to HAL, and a truly intuitive interface experience, we’ll have to stop thinking in apps and screens, and start thinking about how to get computers to talk like a human.

Enter, the chatbot.

Chatbots you say?

You’ve likely heard the term “chatbot” thrown around a lot recently. That’s not without reason. Internet giants and trolls alike have been exploring the emerging technology, and many technologists are calling the chatbot the next wave in UX design.

So what are they?

Chatbots are computer programs designed to interact and simulate conversation with a human. They’re powered by artificial intelligence (AI) so they can respond more relevantly to users.

Like any emerging technology, chatbots live in a spectrum of complexity. On one end, there are chatbots that use a simple set of rules to deliver pre-programmed responses based on keywords. On the other end, advanced chatbots can essentially learn how users speak. They employ natural language processing (NLP) technology to iteratively learn from users and formulate new responses over time.

Well on our way

Regardless of sophistication, all chatbots are designed to make talking to a computer not just simpler, but more akin to talking to a human — a defining departure from user interaction norms in the decades previous.

And while mostly screen and text-based, chatbots are also a departure from app or platform centricity, integrating seamlessly with pre-existing messaging platforms (iMessage, Facebook Messenger, websites). Rather than downloading or “opening” a chatbot separately, all you have to do is send a message to a chatbot through a communication channel you already use to talk to other humans.

Human-like, and accommodating to your natural communication habits: chatbots, while no HAL, are certainly a step in the right direction.

Most significantly, chatbots signal the beginning of a new era of user interaction possibilities — ones that is less and less confined to screens, and more and more integrated with the way humans talk naturally.

That’s not just the science fiction talking.