Making machines more like humans and humans less like machines

Making machines more like humans and humans less like machines

Far from being a declining sector, the rise and rise of contact centres in part reflects the growth of online transactions – and also the increasingly important differentiator that customer care has in defining which product or service is best.

It was therefore with some interest that I read about RBS and its new AI (Artificial Intelligence) system for internal use, called Luvo. This service allows staff at the bank to interrogate a ‘human’ web chat interface, which then returns the answer to their question. How different that is to Google I’m not sure but it’s a bold experiment in trying to make machines more like humans.

There are, however, more tricks that can be adopted to make the automation of customer communication more real. How many times do you contact a service provider for a simple reason only to be asked to give them a round of applause at the end – the ever-present NPS/VOC survey? If you want to be more human then record the fact that the customer has been surveyed within the last three months and don’t survey them again. If you insist, then at the very least vary the question asked. Imagine every time you spoke with your next-door neighbour they asked you to rate the conversation?

It did however give me an idea. When people say rude things in their engagement with automated systems, what you shouldn’t do is say “I’m sorry I don’t understand” or “we couldn’t quite make that out, can you say that again” because that leads to a sweary escalation. Rather accept that the customer is a wee bit upset, let them know someone will contact them. Give them a few minutes to calm down and then use what humans are best at: be nice, calm and sort the problem.

In summary, humans best relate to humans. Make sure when automation is adopted it’s only part of the customer journey – used for the simple bit where a chat is not necessary. But make it dialogue not monologue – so people can reply – and then if the words they say or type trigger a profanity thesaurus (I have over 400 words and terms available in mine, if you’re interested!) make sure that the human kicks right back in. This way humans do the human bit, and computers learn a little bit more about being human.

Dr Mark K. Smith

Founder & CEO