Why the future of customer service is in your face
By David Martin
Chatbots may be fast when it comes to delivering dynamic feedback to lots of users. The best customer service offers an empathetic ear, however, and moreover, it gives the customer signals that concerns are being taken seriously. The secret to getting chatbots to the point where they can fulfill that calling might be right underneath our noses.
“Face is the new interface,” said Mark Walsh, the CEO of Motional.AI. He boldly stated this as if to say that such text and voice bots were already mainstream. He spoke at Humanity.AI — an AI and chatbot conference in San Francisco run by Capital One. The conference’s theme was “People and Bots” and the overarching message was that bots designed with humans in mind can make us smarter, more informed, and more efficient.
Walsh’s point was that face has the potential to be the new user interface because — for human beings — it was the original interface. First, face is better because 55% of communication is visual. The earliest cavemen were entertained — they laughed, learned, and connected by telling stories around fires. “Face time” is important because we check to make sure people’s words match their language as we ask questions such as “Are they being sincere? Are they telling the truth? Can I trust them?” The need to connect and verify that facial expressions match words explains why we continue to have retail-based customer service — e.g. where a segment of the population still walks into their bank branch to financially transact with another human being.
As Walsh’s talk continued, he showed us a video of an animated, healthcare bot interacting with his retired mother via tablet. The bot’s avatar was Pixar-like, enthusiastic, and ready to help — it happily sent a note off to Mom’s doctor asking for a follow up. The bot was non-threatening and almost human even in cartoon form, offering a fun experience to take in that made the audience visibly warm.
But as consumers, what we really want is the benefits of bots — the always-on, the instant gratification — paired with the human connection — the trust, the ability of people to read between the lines, and the scratching of our itch — providing answers and solving problems in a cooperative relationship that lasts.
Walsh’s idea for Motional.AI is to humanize bots with empathy and character a.k.a. “Character Intelligence(TM) Artificial Intelligence” to lay the groundwork for a trusting relationship between consumers and brands. Diving a little deeper, he wants to bridge the uncanny valley, i.e. that intuitive, not-quite-right feeling you get when you look at a robot failing to impersonate a human. In the short term, his goal is to offer brands and their audiences the best face-based bot experience.
Foundations for Better Service
Over the past year, I have observed startups, trends, and research that support Walsh’s vision — seeds of technology that could sprout, grow, and bloom into new face-based customer service experiences. Below are some of the examples that I’ve seen.
Buddy, retirees’ best friend
Truth be told, I have seen Walsh’s demo before in a different context. One of Orange Fab’s startups is a company called GeriJoy, which offers 24/7 geriatric care through a team of remote human caregivers. A canine avatar pops up on a tablet, reminds seniors to take their medicine and exercise, decreases their loneliness through daily interaction, and allows working children to know how their parents are doing.
Two noticeable differences strike me. First, GeriJoy uses an animated dog to deliver care vs. Motional.AI’s human. GeriJoy explains that the avatar offers consistency and having an animal avatar offers some benefits related to pet therapy. Second, Walsh’s demo was pre-recorded, so the depth and scope of the AI or human + AI is unclear, while the GeriJoy dog avatar is explicitly controlled live by trained human beings.
Companies like Microsoft, Kairos and Affectiva are staking claims in the world of emotion detection. These technologies read faces to detect emotions, while a recent article by MIT highlighted early research into detecting emotions using wireless signals based on your heartbeat.
Today, banks such as JPMorgan and Wells Fargo are already using voice detection to fight fraud. Emotion detection will initially be used to measure ad effectiveness and to optimize ads and content. This same emotion detection technology will be able to help customer service agents better manage the their customers’ expectations. Human agents will be able to monitor emotions and mood and make offers to increase satisfaction or cross sell additional products at opportune times.
Optimizing VR-based learning
I first learned of Jeremy Bailenson’s lab at Stanford when Facebook CEO Mark Zuckerberg visited to better understand the potential for social virtual worlds before acquiring Oculus in 2014.
Bailenson runs Stanford University’s Virtual Human Interaction Lab with the stated goals of understanding the nuance of face-to-face interaction and how to use VR to improve everyday life including conservation, empathy, and communication. In one of his projects, social researchers are currently running experiments on virtual classrooms to optimize for learning outcomes. There is an opportunity here to take these learnings from VR in the education sector, fund additional research for mobile interaction, and re-apply them to customer service.
Customer-Driven Customer Service
What happens when we bring it all together? Blending Motional.AI’s Character Intelligence with GeriJoy’s avatar-based geriatric care (driven by a human team), emotion detection, and research on how to optimize customer experience could make our lives better. Here’s how.
The HealthTime app
Imagine that you had a questions that you wanted to ask your doctor, so you launch the HealthTime* app and the image of a human agent pops up on your phone asking, “How can I help?” The service always loads instantly, and no matter whether your question is about your teeth, appointment, diet, exercise plan, vision, bill, current symptoms, or latest lab results, you never have to re-explain your issue or question. The agent is consistent — it’s the same human avatar every time — and he/she is non-threatening and even fun to chat with. For you, it’s seamless. HealthTime never feels like you have been transferred to a different human agent (even if that happens in the background). The avatar is never rude or condescending because the AI literally knows what that experience looks like and can intervene before such behavior appears.
HealthTime may take a minute to research or come back to you with more specific questions, but if the service can’t find the answer shortly, then it calls you back at a convenient time. In fact, the service can tell how you are feeling right now, whether you want to be chatty, cheered up, or you just want to cut to the chase; it senses when you start to get irritated and compensates; and it intentionally ends on a high note delivering a shot of satisfying dopamine right at the end of the call.
HealthTime is a one-stop shop for your health, and the system knows all your medical history. It’s where Kaiser Permanente meets Doctor on Demand on your mobile phone. It’s beautiful because the onus of the experience and communication is on the system and the providers instead of customers. The system offloads the overhead that we face today with managing our healthcare and the associated bills, and free us up to take action.
Now, re-imagine any customer-oriented business this way. Imagine DollarTime for banking, loans, and investment. What about MobileTime, CableTime, and UtilityTime? How can we make customer service proactive instead of reactive? How can we make customer service customer-centric — not managed and driven by the customer? With HealthTime, you manage your health without managing your healthcare.
*Note HealthTime is purely fictional and bears no reference to any real application.
This new system brings up several ethical questions — what’s the top level metric for companies? Should they be optimizing for satisfaction, retention, revenue, or something else? How can enterprises leverage the cumulative rich customer data set to build not only customer personas, but bot personas that optimize customer service based on company goals?
From the customer’s perspective, how personalized should the customer service bot be? If you are a 25-year old Latino man who speaks Spanish and English, then what age, sex, race, and accent should the bot have? Beyond demographics, if you are an impatient extravert who calls up once a month because you might have a basic question, but you are lonely and just want someone to chat or entertain you, then what personality should the bot adopt? Alternatively, if you are a low-margin/high-maintenance customer, should the bot intentionally treat you worse to nudge you towards canceling your service?
Contact me if you are interested in talking more over coffee in San Francisco. I will also be attending Habit Summit 2017 in April and am excited to see Jane McGonigal talk about gamifying life and Buster Benson discuss behavior change.
Disclaimer: The views and opinions expressed in this article belong to the author and do not necessarily reflect the position or views of Orange or Orange Silicon Valley.