The digital assistant has landed – Google’s most recent release of Duplex, an AI equipped digital assistant capable of making natural language phone calls, is a lightning rod in the industry. The new advancement suggests humans will soon be equipped with personal bots to conduct transactional conversation to schedule dinner reservations, hire plumbers, and order pizzas. Ignore for a moment the strong implications that we are co-opting humans with robots, and consider the intelligence itself.
The goal of the Google Duplex is to “get things done,” and as pointed out in a recent presentation by the company – making a phone call is still a critical part of this process. Through the application of text-to-speech and deep neural networks, it appears that Duplex has achieved the ability to converse with simple speech by building an AI equipped digital assistant that mimics human-like vocal tones, feedback, and adaptability. The system even built in techniques to mimic the natural pauses, “umm’s”, “ahh’s”, and other speech disfluencies that characterize everyday conversation.
Before we jump on the bot bandwagon though, let’s pause and take a moment to consider the full scope of conversation. Simple conversations that involve making appointments or reservations (as described in the Google Duplex blog), are limited in scope. As Google stated when introducing the AI assistant, “Duplex can only carry out natural conversations after being deeply trained…in narrow domains. It cannot carry out general conversations.” So, contrary to replacing all customer service providers, Duplex may just represent the evolution of the virtual chatbot – in adding some human qualities to transactional services. Furthermore, is this level of “uncanny” human-like abilities something we should look forward to? Do customers want to interact with a “human-like” bot capable of limited transactions that are narrow in scope and/or depth? Or are customers that pick up the phone looking to speak with a person who can understand the complex contextual aspects of their problem?
The Complexity in Service-Oriented Conversations
Think of a time when you are meeting a colleague for the first time via phone, or call customer service at an insurance provider to help you find a new surgeon, or it’s winter and your roof is leaking due to the heavy falling snow. All of these scenarios encompass subtleties and circumstances not seen in the transactional conversations displayed by Duplex.
Often times, beyond service, these conversations require connection. Simple conversations can feel shallow because, for most of us, without additional conversational cues that share insight into a person’s state of mind when speaking, it can be hard to read the person and fully understand the meaning behind his or her words. Indeed, in a conversation where we need more than an appointment, we need to establish a rapport, and in some cases, empathetic relationships. In other words, when conversing with chatbots we miss out on that satisfying feeling of a good conversation – one in which both parties feel understood and like they could put themselves in the other person’s shoes.
This suggests that the demand for customer service agents will continue to grow over the next several years, as they handle more complex interactions where dynamic, broad conversations are required. In fact, a recent Pew study focusing on 21st century employment called out the importance of flexibility in the workplace moving forward. “The tough-to-teach intangible skills, capabilities and attributes such as emotional intelligence, curiosity, creativity, adaptability, resilience and critical thinking will be most highly valued” in the 21st century.
Why Should You Care about Conversation Quality?
Conversations matter beyond day-to-day family life and friendships – they are also crucial in the workplace. For instance, customer service jobs in the United States, which require difficult and often emotional conversations, are the fastest growing workforce in the U.S. economy. Research suggests this customer service space will increase by 25-36% over the next ten years. These roles require collaboration and engagement between customers and vendors. Whether it’s a conversation between colleagues, between the customer and vendor, or between neighbors, our ability to listen and respond is strongly tied to how successful we perceive the outcome to be. So instead of focusing on more human-like robots, perhaps we should use technology to help augment humans to have better conversations.
How Tech Is Improving Our Ability to Empathize
Professional service industries, including the customer service space, are adopting new technologies as “guideposts,” which can augment workers to improve conversations. In the same way that human interactions improve with quality conversations, our relationship with technology may also generate more positive outcomes and greater practice of empathy in communication.
Innovations in machine learning are specifically focused on emotional intelligence and looking to help us retrain our brains to identify moments for empathy. Among the fastest growing sectors in the U.S. economy are computer systems and professional services. These two sectors intersect at the customer contact center where companies are rapidly incorporating digital technology to augment their customer service workforce. Although Google’s Duplex may offer bots who sound like humans, as a society, let’s start with humans and cultivate the skills needed to deliver personal connections in all conversations.