Talking to robots and why the future of communications is already here
When you imagine talking to a robot, you probably imagine a humanoid form, like C3PO or Honda’s Asimo. You may not have imagined that robots – often existing purely as code running on a server – are already playing a role in customer communications.
These kinds of everyday interaction may seem a little mundane for cutting-edge technology, but in reality this kind of large-scale, repetitive task is the ideal learning environment for artificial intelligence. Tasks like directing customers to online resources, handling customer queries and sifting through thousands of emails are precisely the kind of high-cost, low-value task that can easily overwhelm customer service agents and produce a poor customer experience. But organisations wanting to make these tasks more efficient must tread carefully, ensuring that any initiative to reduce the administrative burden never creates stress for the customer, but only makes their lives easier and saves them time.
Robots rule at Ocado
In this light, it’s easy to understand why organisations are turning to robots to meet these dual needs. At Ocado, the grocery delivery company, they implemented a machine learning system to help them manage seasonal fluctuations in the number of emails they receive. At certain times of the year Ocado get more emails from customers wondering when their delivery will arrive. One solution would be to hire more customer service agents to respond to these emails, a costly process that would involve recruitment, training and on-boarding a temporary team to deal with a predictable problem.
Instead, Ocado developed a machine learning solution based on Google’s TensorFlow, which uses natural language processing and sentiment analysis to sort through incoming messages, organising emails into different categories based on their content. This reduces the burden on human staff, giving customer service operatives the ability to quickly support the people that need it most – while side-stepping the laborious task of filtering inappropriate emails.
In Sweden, Swedbank has implemented an artificial intelligence called Nina to answer customer queries. Developers originally populated Nina with 100 of the most common queries, but since her installation she has quickly learned from thousands of interactions. Nina can identify the nature of a query and provide a link to online help content, or ask the customer to call or email for a more detailed response. Once again, having a robot on the front line of customer service means that simple queries can be resolved quickly, giving human agents more time to deal with complex requests.
Automating complex validation processes
Robots can be trusted with more complex interactions too. For example, car insurance companies can reduce the burden of checking that customers have a valid licence registered at the correct address by using an electronic system that knows which customers have not had their licence checked. Those customers receive a message on their mobile, asking them for evidence. If the customer chooses to proceed, the app takes control of the camera and the flash, and then gives the customer on-screen advice on taking a clear picture of their licence.
The system can then extract the data required from the image, passing the data back to the solution so that the customer’s record shows they have had their licence validated. This single process, which might previously have involved several team members, and considerable inconvenience for the customer, can now be automated and completed in seconds.
At DocCentrics, this is exactly the kind of process that we're automating with robotics. We're helping our clients to simplify and accelerate their customer communications, while also improving the experience for the customer. Our Dynamic Communications service creates individual, bespoke communications that are effectively microsites, dynamically generated based on the customer's history and preferences. Dynamic Communications can integrate a wide range of additional features, such as downloadable documents and digital signatures, so the customer can manage everything from a single portal that works on all their devices.
Is this the droid you're looking for?
If you’re still looking for a robot that’s more like C3PO, you’ll be glad to hear that in Japan, the Henn-na Hotel is staffed entirely by robots – although one of the receptionists is a velociraptor wearing a bow-tie. And in Virginia, the Hilton McLean features a small humanoid robot called Connie. Connie is a NAO robot from Aldebaran Robotics, and she uses IBM’s Watson AI to respond to a wide range of customer queries. Watson famously won the game show Jeopardy in 2011. Watson was also invited to a Q&A by the Oxford Union.
Avoid the uncanny valley
While these examples show how robotics can simplify organisational processes and improve the customer experience, questions remain about how the public will respond to increasingly overt interventions from non-human brains. Will our digital assistants feel magical, or creepy? The 'uncanny valley' is the point at which human replicas become almost convincing - but clearly not quite human. It's at this point where robots are most likely to cause revulsion in observers - clearly not the kind of response most organisations want to elicit in customers.
Many of us have already experienced the slightly unsettling sensation that Google knows us too well when our phone suggests a quicker route home, when we’ve never explicitly stated where we work – or live. So while it’s great to have technology making our lives more joyful, or efficient, organisations must remember that AI can’t currently match the empathy of human interaction – and being too smart can result in customers feeling watched.
So while we're a long way from having robot helpers in every home, digital assistants and AI are already working hard, behind the scenes, to enhance our communications and interactions.