What does AI mean for customer service centres?
As a customer fighting the now traditional battle to get a promised callback out of Vodafone, I’m rather looking forward to artificial intelligence (AI) taking over a few jobs, smoothing a few processes and actually keeping promises. After all, a bot doesn’t get a bad case of the can’t-be-bothereds on a Friday afternoon.
But how far will AI change the way customer service works? And, if you’re working in a call centre or a customer service centre, what does it mean for your job?
The answer is ‘it depends’.
Back in the late 1990s we began using neural networks to predict statistical outcomes from huge data sets. We started seeing the first glimmers of predictive customer behaviour. Now, similar principles – but far more sophisticated technology and much more data – means a bot can look at your chat question and give you several relevant answers.
And that’s the challenge. AI is still at a point where it’s great at giving answers to questions with a relatively limited range. That makes big demands on customers. It means they need to ask the right question in the right way to get the right answer. In regulated and technical industries with their own dense, internal jargon, that can be a real challenge.
A good human agent can often pick up on what a customer wants even when their question is vague, poorly expressed or just plain complicated. Current AI is clever enough to contextualise a question but not quite sharp enough – just yet – to sift meaning from an ill-defined question. It’s still garbage in, garbage out.
It’s early days, and as AI gets better at predicting answers from larger and larger datasets it’ll get better at giving the right answer. But at the moment “AI is not good at answering the type of questions that require an ability to grasp meanings across a broad spectrum,” says Noriko Arai, a professor at the National Institute of Informatics.
Where AI, portals and human beings work best.
You can add to that “AI is not good at dealing with customer emotions”. And that’s the critical bit. Because so much customer interaction (and experience) is ostensibly about process but is really about emotion. Remember the last time you called your bank and the agent did everything by the book, it all worked and yet you came away feeling disappointed that they weren’t really engaged?
In fact, an AI agent would give you just the same feeling. That’s because emotions, to have any impact, must be genuine. And to be genuine, there has to be a human involved. A computer-generated “how are you today?” no matter how apparently sincere, can never be more than fake – even more so than ‘missing you already’.
AI systems will change the way our customer service centres interact with clients. They’re already beginning to prove their worth in taking away tasks from agents in the form of knowledge bases, instructional, process-heavy and transactional interactions. But for anything involving customer emotions, AI is out of its depth – and always will be – simply because of its very artificiality.
So what to do about AI in customer service? Put it to work where it does most good, in areas where customers need instructions, speeding up transactions and ensuring consistency. But leave your teams to do what they do best – behave like human beings helping other human beings.