While chatbot and conversational artificial intelligence (AI) technology is now found in a wide range of consumer solutions—smart speakers, smart TVs, various apps and software experiences—many advances in the technology continue to be driven by the sales and help desk markets. You see it anywhere a business might touch a customer, from digital brand ambassadors and AI sales specialists to virtual support staff that look and sound remarkably human.
With the pandemic turning video conferencing technology into a primary communication channel, a chatbot’s next evolutionary step must address not just how natural it sounds but how natural it looks. That’s where companies like UneeQ are looking to make their marks.
“UneeQ Digital Humans is an AI company that focuses on solving for the synthetic behavior and animation that surrounds conversational AI,” says Danny Tomsett, CEO of UneeQ. More than just avatars or advanced caricatures, the company wants to make interacting with conversational agents more natural and realistic.
“A lot of our customers create brand new characters, which are digital humans designed to help with a range of problems on a customer journey. That could be helping them make a purchasing decision or complete an application or a form, or just help with various frequently asked questions,” explains Tomsett. “We now have over 1,000 companies building digital humans on our platform.”
For example, one UneeQ digital human, Sophie, was built as a free public health advisor that focuses on helping users with COVID-19 queries but initiates the conversation as you might a normal video call.
A recent survey from Indiana University on COVID-19 screening sessions with 371 participants found that persons viewed interactions with chatbots or voice assistants more positively than interactions with human agents. Most especially when it came to disclosing personal or health-related details. Digital human solutions, like Sophie, can be deployed in a variety of ways—on the web or in mobile apps, as well as kiosks in pharmacies or other public areas. When used efficiently, digital humans can help cut long lines of patients seeking information, or even triage mild from serious medical calls before handing them off to human health providers.
How Digital Humans Work
To help me understand the digital human experience from a business’ perspective, UneeQ gave me access to its toolset. For those who want to try it for themselves, it’s a Software-as-a-Service (SaaS) solution with a free trial available via the website.
Once you’ve got access, you select from a dozen male or female digital human faces. The agents look ambiguous in terms of skin color and racial features, which UneeQ says helps suppress biases. Even out of the gate, I was delighted by the nuance in their facial movements and seamless speech. My initial reaction was that they seemed helpful and non-threatening. I certainly didn’t feel any “Uncanny Valley” triggers, and neither did I feel dismissive towards the digital humans like I would with less detailed and more cartoony avatars or even an old-style telephone IVR system.
After selecting your digital human, you name it and then select an appropriate background image. You also need to select the corresponding voice the digital human uses, and there are various choices depending on language and region. The final step is designating which AI back-end system will serve as the AI’s operating system.
Once set up with the corresponding AI back-end, digital humans will listen to your question, pass that to the solution platforms or knowledge base, gather the answer, and finally repeat it to the customer.
“So, as a quick example, ‘Congratulations! You’ve been approved for this home!’ Is a very positive output. And the digital human would understand it as positive, just like you and I would, and would say that with a level of enthusiasm,” says Tomsett.
(The above video shows Rugby icon Sir John Kirwan with DJK, a cloned digital human that serves as a mental health coach for health provider MentemiaMentemia while emulating a familiar face and voice.)
Digital humans work by using both verbal and non-verbal cues, like tone of voice and facial expressions, which human users pick up quickly and naturally. This is done by extending the AI so that it can determine behavior and emotional quotient (EQ) and have that information show up graphically as well as through its speech. The AI directly reflects these values on the face of the digital human and the results are impressive.
The agents we created, even using the basic trial tool set, felt surprisingly alive because they seemed emotionally engaged with the conversation. Whether it was the virtual wind on their hair, subtle facial changes to represent breathing, or eye movements used to emote thought and feelings, there was always a sense of connection that made any interaction seem more authentic. Given sufficient time to train such an AI for an actual conversation, we found it easy to imagine a successful dialogue.
Users can think of digital humans as the front end of an application that handles a specific kind of query or transaction. These solutions are merely shells without the cloud access and the various knowledge bases that power their interactions. Once an UneeQ customer, businesses can work with the company to customize their AI back-end for whatever use case they require.
“Major technology companies provide the back-end platforms that drive the digital humans. So some that you’ll be familiar with be like IBM Watson, Google Dialogueflow, Microsoft Bot Framework, and Amazon Lex. These are all readily available to any business. And in most cases, they start with a free offering, So, for small business that can be actually very affordable,” Tomsett says.
The Next Step of Conversational AI
UneeQ sees its digital humans as the next evolutionary step of AI-powered chatbots and conversational AI. That conversational technology can access a host of support databases and solutions is nothing new, especially for those of us who use voice assistants and support chatbots online. Where UneeQ seeks to add value is by adding the human touch, so to speak.
“Over the past 10 years, we’ve seen a lot of innovation in digital self-service solutions and automation. Chatbots and voice assistants have proven themselves effective at transactional business solutions,” Tomsett said. “Digital humans are considerably more valuable than any alternative [since] they aren’t just transactional, they are “interactional.” They provide the scale of digital, but can bring to life a personality that helps users navigate their digital journey in more meaningful ways.”
As mentioned, once customers have engaged with UneeQ, they’ll have access to its back-end platform, which enables any user to customize the digital human’s AI engine with training data. That data would include not just the types of questions it might be asked and the responses it needs to provide, but also the appropriate emotional intent those answers should have.
“That process used to take quite a while,” Tomsett recalled. But he points out that back-end AI companies, like above-mentioned Amazon and IBM, have poured billions into the technology of the past several years resulting in significant improvements around AI training and implementation.
“The time to value has been reduced considerably,” he says. “[We’re down] to weeks, not even months, to have a really effective conversation.” Tomsett feels this can significantly lower conversational AI’s barrier to entry, especially for small and midsized organizations, an opinion with which other experts agree.
Gartner has predicted AI will make up 95 percent of common customer interactions with businesses by 2025. And Tomsett puts weight behind that opinion by showing off other digital human examples aside from Sophie the health advisor, including Mentemia, a mental health coach, and Mia, an interactive bank agent.
