The American health-insurance company Humana has, by the help of artificial intelligence, developed a software that is trained to detect conversational cues, in order to train call-centre agents and supervisors when the conversation is happening, and a customer becomes frustrated, Wall Street Journal reports.
The AI tools applied are supplied by Cogito Corp. and they are able to detect emotions such as agitation by “listening” for signs such as the pitch of the customer’s voice steady rising, or whenever a customer and an agent is talking over each other. Cogito Dialog, as the app is called, will then notify the agent by sending an electronic alert outlining the “issue”, such as “speaking quickly”, “sounds tense” and “frequent overlaps”, to enable the agent to change their technique.
As previously mentioned it is essential that anyone working with dealing with clients over the phone has a warm and welcoming voice. This is even more essential to Humana, Geeta Wilson, Director of Consumer Experience explained to WSJ, as they are a health-care company dealing with clients queries regarding procedures and mediation. “A health-care company has to have empathy and compassion,” Ms. Wilson said.
It is a growing trend that researchers and technology companies are now using natural language processing to study how these systems can detect conversational cues, where one of the goals is to establish a more efficient rapport with the users. It is also very helpful that artificial intelligence tools are now becoming more affordable.
Ms. Wilson explained that Humana’s goal is to predict customer behaviour. “We want to know what the consumer is going to do before they do it, so it feels like we know them,” she said.
So how has Humana developed their system? Data from 100 000 customers service calls was fed into the AI system earlier this year, to analyse these key cues, such as pauses, keywords and interruptions, and then associating them with any outcomes of the interaction. This has resulted in these ‘digital coaches’ that can listen for stress etc., and then instantly send suggestions to the agent to either talk less, slower, moderate his/her tone, etc.
“Those alerts help agents determine how they should change the conversation to make sure the member feels like they’re being heard.” Ms. Wilson said.
This will hopefully be very helpful for both customers and the agents, as it will potentially maintain a calm and collected tone throughout the conversation.
This article was originally found at: http://blogs.wsj.com/cio/2016/10/27/artificial-intelligence-helps-humana-avoid-call-center-meltdowns/
Photo Credit: Google