It’s not that chatbots have a failure to communicate. They just have to do it with feeling.
For the banking executives overseeing the development of virtual customer service, the hard part is making sure that these digital assistants not only can help a client with a transaction, but that they can relate to them better.
"They are getting better" at empathy, said Jerry Gupta, senior vice president at Swiss Re. "But they are not there yet.”
Gupta spoke
Empathy and the setting of expectations are crucial in creating a great overall customer experience, Gupta said. Many companies aspire to have AI-driven processes handle 80% of routine customer interactions, and let human agents troubleshoot complex problems. But even handling relatively straightforward questions, chatbots will need to detect emotion during these queries and act accordingly, he said.
Aiding developers is the ability of machine learning, said Steven Ramirez, CEO of Berkeley, Calif.-based data science firm Beyond the Arc. Virtual assistants are trainable, he said. Just like human staff, they can be taught how to deal effectively with customers and use the right words.
That will form the basis of the ability to build trust between a customer and a bank, Ramirez predicted. “You are going to see an empathy bot.”
In addition to learning how to empathize, effective bots will understand different regional dialects, the generational divide in conversation, and understand Twitter slang and emojis, said Ryan Miller, senior vice president and design and delivery team leader at Wells Fargo’s Innovation Group. The response from the bot must be consistent with the brand’s voice, he added.
Another challenge that AI banking teams will need to confront is context, Miller said. If a client is speaking with a banker, that person can assess the environment and quickly deduce what information to relay publicly. AI-driven chatbots will need to have this contextual skill. For instance, if a customer asks for his or her balance in a car with other people, the bot will need to quickly discern that it should relay that information in general terms, Miller said.
With the explosive growth of AI and different channels for customer engagement, bank executives are also confronting issues tied to keeping a consistent brand identity.
Miller said Wells Fargo is looking at how to control customer experience on channels built “off-property” versus channels built in-house, where consistency is insured. Other issues include whether to give bots names or a persona, and how do these considerations impact customer perception.
Examples of how AI is improving customer experience were showcased at the event.
In one case study, BMO Financial showed off its Facebook messenger chatbot, BMO Bolt, which was released earlier this year along with the bank’s Twitter chatbot, BMO Virtual Assistant. Backed by machine learning, BMO inputted into Bolt the 250 top customer questions that the bank receives from all its customer feedback touch points. If the bot could not answer the question, which happened 45% of the time, the query would be routed to a human, said Sumit Sarkar, head of customer experience and strategy for personal and business banking marketing at BMO.
There is a ready market for these AI-driven applications, said Arvy Rajasekaran, chief information officer of digital channel technology for Ally Financial. In the past year, Ally rolled out its Alexa skill, which not only gives customers their balance, but also transfers money and converts the price of an item on sale into working hours.
“The biggies surprise has been level of adoption. The adoption is growing,” said Rajasekaran, who did not go into specific numbers. “It speaks to how well the conversations are designed, how well the analytics work.”
What’s also surprising is that clients age 45 and older are the fastest growing demographic embracing conversational technology, at least in the case of TD Ameritrade’s Alexa skill, said Sunayna Tuteja, head of strategic partnerships and emerging technology at TD Ameritrade.
“They are strapped for time. These technologies are becoming part of their lives, and they are ubiquitous,” she said.
With all these breathless developments in AI and banking, Ramirez cautioned that financial services need to make sure these new applications are not simply window dressing.
“I am really excited that all the power of machine learning is really now laser-focused on customer experience. That’s a really long time coming,” he said. “But I think for banks, what I see is a major challenge, particularly bank IT departments, they start with the feeling, ‘We are good at data, [so] we are excellent at technology.’
“I think banks are excellent at standing up infrastructure. They are good at aggregating data and creating metadata, which are all fantastic. But I see this huge challenge in” transitioning from “an archiver of data to an active user of data to drive machine learning and actually make decisions. That’s where it breaks down. Sometimes, you think you are doing it, but really you are arranging the deck chairs."
