COVER STORY
Rise of the bots Expect chatbots to pick up more of the grunt work, but there’ll still be a need for human advice. By Angela Cuming
T
he next decade will mark significant changes in the insurance industry, with insurers working overtime to rebuild trust and keep customers happy. The key to succeeding in those goals may all come down to replacing call centre staff with fully automated chatbots, one commentator says. Chatbots are computer programs or simply artificial intelligence (AI) systems that can conduct natural-sounding conversations with humans. Due to rapid growth in technology, chatbots have recently made great strides in the insurance industry and before too long will be a common feature, say experts like Dr Michael Naylor, a senior lecturer at Massey University. “By the end of this decade if an insurance company isn’t using chatbots that company will cease to exist,” he says. Chatbots are perfectly aligned to the insurance industry because they can provide customers with efficient service when responding to quick and common requests, such as passwords, policy copies, and billing questions. Chatbots can speak to and understand people to a degree that feels nearly human, allowing them to personalise and automate multiple processes and enhance the relationship between the insurer and the policy-holder.
“Advances in technology mean that customers would not be able to tell that they were not speaking a person on the other end of the phone,” Naylor said. “It means an insurance company will be able to handle roughly 95% of customer inquiries without actual staff.” So, what would that mean for insurance companies? Lower operating costs for one thing. “One way insurance companies are going to be able to survive in an increasingly competitive market is to have less staff and slowly reduce the firm down to a software base,” Naylor said. And while chatbots were expensive to set up the cost was more than offset by their long-term benefits, he said. He points to what happened after the 2011 Christchurch earthquake as an example of where chatbots could have been of use. “One of the main problems with Christchurch both for the insurance companies and the EQC was the sheer amount of work they had to do versus the amount of work they actually could do,” Naylor said. 12
March 2020
“It was a case where EQC had 20 staff, mainly financial people, and how do you deal with 40,000 upset and stressed customers all at the one time? You can hire extra people but they have to be trained up properly, but chatbot software can cope with one person or 10,000 people, it makes no difference, even in times of natural disasters the likes of which New Zealand is no stranger to.” Would a customer know they were not speaking to an actual person on the other end of the phone line?
PEOPLE THINK ABOUT A COMPUTER BEING MONOTONE AND WITH WORDS EVENLY SPACED AND THE LIKE, SO THAT’S WHAT I MEAN BY IMPERFECTIONS. WHEN YOU SPEAK TO A CHATBOT THERE’S NOT THAT ROBOTIC VOICE
Naylor said they would have no idea. “There’s something called a Turing Test, which means can you tell if you are talking to a computer, and in terms of voice codes no you can’t,” he said. “I mean when you think about it people think about a computer being monotone and with words evenly spaced and the like, so that’s what I mean by imperfections. When you speak to a chatbot there’s not that robotic voice.” That’s not to say chatbots will be infallible and never make a mistake. “Yes, they will make mistakes but if you have a worldwide firm, once
