When I started at university in 1999 (grandpa shuffles along with his walker), most of the effort in the ‘computational linguistics’ department went into search engines, speech to text, and text to speech projects. All of which are heavily based on the rules of grammar and sheer statistics. Results were buggy, error rates were high and every computer sounded like Stephen Hawking.
Recently, well relatively recent, Apple launched the Dutch version of Siri. I was in my car and the radio DJ did a little test. On the radio, he said in Dutch: “Hey Siri, set an alarm for 5 am”. Next to me on the passenger seat, my screen lit up and Siri had actually set the alarm! I pressed the home button and told Siri to cancel it. Then I called Siri a Jerk for setting the alarm, and he responded he did not like it when people spoke to him that way. Really…! And then it occurred to me: what we were doing one and a half decades ago is now apparently mature enough to be installed on an iPhone (and you know Steve only wanted the best for us humans). My phone was interpreting what I was saying and naturally speaking to me. Very 2015-ish.
Of course, interacting with a phone is relatively easy, compared to fully understanding a language. The phone scans for a small set of commands and acts on it. But with new small devices such as smartwatches, speech will become a much more prominent interface. And we are not nearly ready for that.
What if you would be able to go to the website of your favourite airline and then just have a normal conversation with it, telling the site what you want, getting results and book a flight? Maybe even without touching a keyboard and not having a screen in sight. What would search results look like if not on a screen but delivered by speech? We would have to be much chunkier to prevent memory overloads. But we cannot make the process too long as we would lose cohesion.
Search queries will become more general. Have you ever tried to tell Siri what you wanted in one sentence? The more tasks you put into one sentence, the more you will probably twist your tongue. In a natural conversation, the available information will build up over time. An assistant should be able to filter by asking relevant questions. But filtering could also be done in other ways. When visiting a travel agent, the agent will probably filter using personal experiences, feedback from customers, and parameters like special offers.
And that is where things get really exciting. We can expect more hybrid interfaces. An interface that interacts by voice, shows you results, and active filters on a screen that will build up during the conversation. Which you can alter using touch or mouse/keyboard interaction. For a travel agency, we analysed the conversations employees had with their customers. They take a very different route than classic browsing. Normally, after getting the basic requirements, employees show possible destinations quite early in the conversation. They use the response from the customer as a filtering tool (no, I do not want a big resort, too many people, etc). So actually, during the conversation, the total set of possibilities is narrowed down using positive and negative filters by using examples. In a hybrid interface, you could mimic such a conversation.
The trend in the following years will also be to integrate the service element into webshops. In a regular shop, service and products are intertwined. You browse products and an employee may give you advice. On a website, it is all purely functional. What if, during browsing, you could have a conversation with a robot that can alter search results or set you on a different track? “What about this product?”, “Did you know we have this in green as well? Let me show you…”. During the conversation the bot learns from the user, becoming more and more relevant.
I am truly excited we are now entering a new era where we can put the techniques we were experimenting with one and a half decades ago to work. We can now focus on how to make new interfaces usable for the grand public, greatly widening the field of interaction design, user experience, and analytics. Let the experiments begin!
