Why artificial intelligence needs a human touch

19 Jun 2018

We’re halfway through 2018 and I can say with confidence that the big tech buzzword of the year is artificial intelligence (it’s to 2018 what blockchain was to 2017). My LinkedIn feed is flooded with AI stories. It was also a major topic at the recent GREAT Festival of Innovation held in Hong Kong, which saw a range of innovators and thought leaders across sector and industry come together to discuss the future of how we will work, live, play and learn.

People are talking about AI a lot. I like to think that in 2038 our AI overlords will be talking about everyone talking about it in 2018 – a reason I’m going to say kind things in this post about our future masters.

AI will lead to major change, and probably more than we think. But for now it’s still in its infancy. In its most basic form, we have already seen many companies build chatbots that perform simple tasks like finding flights and communicating bus times.

Gartner predicted that by 2020, customers will manage 85 percent of their relationship with an enterprise without interacting with a human. A separate survey by Oracle claims that 80 percent of brands will be using chatbots for customer interactions by the same year.

However, building an effective customer service bot is easier said than done. Technology is only part of it. A good bot needs a personality, and an engaging one at that. When shopping in my home country, one of the reasons I go straight to a human cashier instead of a self-checkout is to talk about the weather as I buy my milk. As a rule, humans don’t like talking to machines (sorry, overlords).

Keep it real
Way back in 1997, when the PalmPilot was cool, software entrepreneur Harry Gottlieb wrote a white paper in which he shared a set of principles for creating the illusion of awareness. One such principle was to use dialogue conveying a sense of intimacy.

Indeed, recent evidence shows that people favour faults, errors and awkwardness in machines.

While true AI is still the realm of science fiction, from Her to Passengers and Blade Runner 2049, a bit of basic banter can be achieved today – something that leaves us amused as well as informed.

Facebook Messenger bot Bus Uncle, a virtual Singaporean bus driver, is a good example of this. I find “his” use of Singlish engrossing (I don’t live in Singapore, so it has no use to me otherwise). Described as “grumpy, friendly, and helpful”, Bus Uncle communicates in a tone that many commuters can readily identify with.

Enter the PR professional
To make an AI relatable, its design should not be left solely in the hands of a developer or a UX specialist. Storytellers should be considered instead to give AI a semblance of humanity.

The big tech companies, always a step ahead, have recognised this for a number of years. Google hired a Pixar story artist to give Google Home a personality, while Microsoft lured a Hollywood screenwriting veteran to bring its virtual assistant Cortana online.

PR practitioners, of course, are natural storytellers. We might not be churning out Hollywood scripts, but we know how to craft messages that are tailored to different audiences.

As a Brit, for example, I would rather be addressed by a bot in familiar language if I need UK information or services. So instead of “Have a good day, Sir” we have “Cheers, mate”.

It also means injecting humour, emotion, emojis and gifs in a bot that might otherwise be perceived as dry and, well, robotic. It’s called “chatbot” for a reason – you would want to chat with it.

One day, AI will be extremely good at all of this – from detecting emotion to responding in way that passes the Turing test with distinction. But until that day comes, let’s get people involved.

To make an AI relatable, its design should not be left solely in the hands of a developer or a UX specialist. Storytellers should be considered instead to give AI a semblance of humanity

Let's stay in touch Let's Stay in touchLet's Stay in touch