NTT DATA Business Solutions
Thomas Nørmark | May 26, 2020

How Can Digital Humans Improve Customer Service?

Digital humans add an emotional quality to human-technology interaction.

(6 min read)

The last decade has seen chatbots become a standard part of the online customer experience. Thomas Normark, the Global Head of AI & Robotics, Innovation Technologies at NTT DATA Business Solutions, explains how digital humans, the next generation of chatbots, will transform the way people communicate with technology and why they are lifting customer service to the next level.


When we developed Kia Mia, it was not only a cool experiment in how chatbot technology could be applied in a physical space, benefiting colleagues on the sales floor at a Kia Motors dealership in Denmark. It was a proof of concept that showed the potential for digital humans to change how customers experience retail and hospitality, taking a glimpse into the future of customer service.

In a telephone interview, which has been lightly edited for clarity, Thomas Normark explains why the technology is important, and where it’s headed next.


From Chatbots to Digital Humans via AI

Tell us a bit about the history of chatbots. Online, they’re everywhere. Why is that?

Thomas Normark:

Kia Mia, a digital human, is an example of how technology is set to transform the customer experience.

We are now interacting with technology on human terms instead of on technology’s terms. In the beginning, we had to type command prompts and stuff like that. That developed to the point where we could type, click, drag, and drop. These are abstraction layers that help us communicate with technology, but they are on technology’s terms.

What we have done in recent years – thanks to advances in technology like deep learning and natural language processing* – is begin communicating with machines in our own language.

This is a major shift, and it started with the chatbots, then evolved to service bots like Pernille. The next generation, which we have not seen a lot of yet, is what we call the digital human generation, where we give the machines a human identity, a human look and feel, to try and further humanize the interface. That’s what we’ve done with Kia Mia.

* Natural language processing is where linguistics, computer science, and artificial intelligence come together to allow machines to hear and understand human speech and to use it themselves. Kia Mia can speak 120 languages!


Customers Get in Touch with Digital Humans

Besides the human look and feel, how else does this next generation change how we interact with technology?

Chatbots have allowed for a conversational interface between technology and users, but those users typically sit away from the service, at home on a laptop, for example, or use a telephone. They’re not present in front of a product.

When we talk about digital humans, we’re talking more about the physical space. We can place the screens where these digital humans are in a car dealership, in a retail store, a library, or a tourist information center.

They can also do things we can’t do with a self-service portal, or a chat bot, because they can communicate emotions. And that’s one of the most important aspects of this new generation, the emotional connection.


Digital Customer Service Becomes Emotional

What benefit does that emotional connection bring?

There have been several studies over the last few years that show that emotionally connected customers are more than twice as valuable as highly satisfied customers. They will come back again and again, so their lifetime value is higher.

So the question is, how can we create that emotional connection? It turns out that more than 90 percent of emotional communication is done nonverbally. It’s about body language, micro expressions, and emotional feedback. All of that lets me know that you understand what I am saying. These non-verbal things are so important.


Kia Mia, the Digital Human in Real-Life


Digital Humans Learn and Constantly Improve

How does machine learning fit into this?

This is only possible because of machine learning. We could only learn to create photo-realistic humans by creating a machine that could look at real humans and try to replicate what it saw. So just the creation of the avatar is based on machine learning.

Another example is the emotional feedback. Kia Mia has a camera that works as the eyes of the avatar. We trained it to understand emotions, so it can see if you’re disgusted, surprised, happy, or angry. It has learned all these moods that we have as human beings.

The way we did that is by giving it thousands and thousands of examples of humans being satisfied or happy or surprised. And over time it learned facial expressions. The more it learns, the more it experiences, the better it will be.


Digital Human’s Evergrowing Knowledge Base

Besides the emotional connection and being at a physical point of sale, what added value do these digital humans bring?

One factor is scalability. We could produce hundreds or thousands of Kia Mias, if we needed them.

Another is the knowledge base. They can all see, they can all hear, they can all understand, they can all talk, and they can all remember. But the knowledge base varies from project to project.

They have a huge depth of knowledge that is very specific. In the case of Kia Mia, it knows everything about Kia Motors’ cars, but it might not know anything about tourism in Copenhagen, for example.


How does one build that knowledge base?

There are two parts. The first part is static information, which does not change over time. This part is similar to a traditional chatbot, where we would hard code questions and answers. A good example would be the opening hours of your store, which probably won’t change much. And if they do, you could change them in some sort of configuration panel.

The second part is what we call dynamic information, which is information that changes all the time. This is like the weather or a dynamic price. And for collecting the dynamic information we use something called Retrieve and Rank. The AI will use the internet to try to find the best answer.

In the Kia case, we check the weather. If there’s ice on the road, or if it’s slippery from rain, we’ll check that and tell the customer, “Take care. It could be slippery when you drive out of this driveway.”


Natural Customer Interaction

When Kia Mia went live, what about the customer interactions surprised you?

Kia Mia, a digital human, is an example of how technology is set to transform the customer experience.

The first weekend when we released it in a car dealership, we learned a lot. One thing was that people will have to get used to this new mode of communication.

At first, they didn’t really pay attention to Kia Mia. Then we put her in a new place where they couldn’t miss her, and then people began to interact with her. So, we really had to nudge people toward this new service.


Did different groups of people interact with Kia Mia differently?

The thing we were most afraid of was that people would respond negatively. Maybe they would find her freaky or unnatural to talk with. Fortunately, people found it very natural and a cool as well as memorable experience.

Elderly people asked more simple questions, whereas younger people were much more natural with their conversations. But they are used to talking with Siri or Alexa or Google Home.


The Future of Digital Humans

After the success of Kia Mia, where do you see this digital human technology going next?

We have created this digital human platform, we call it the it.human platform, and in that platform you can create digital human personas. We’re launching it in May, and it will be globally available.

Kia Mia, a digital human, is an example of how technology is set to transform the customer experience.

One persona could be a digital human receptionist, who could stand in an office, greeting the guests or the colleagues, telling them about parking rules or fire escape rules, and greeting people with whom you have a meeting. All the stuff that a normal receptionist might do. But that’s just one persona. We have a digital sales clerk, we have a digital concierge, or a digital tour guide. As of right now, they speak 120 different languages.

We see retail, hospitality, and the public sector as the three industries where we think the technology is most applicable.


What’s the next step or development that excites you the most?

Right now, these digital humans are on a screen. The next step would be to project them as holograms, so they move out of the physical hardware and are even more alive. If we can combine hologram technology with the digital humans, we’ll have something both spooky and amazing.


Improving Customer Service with Avatar Technology


All Infos about the Kia Mia Innovation Case

Improve Your Customer Service – Get New Insights

Click Here and Level Up Your Customer Experience