Satya Nadella, the CEO of Microsoft, notwithstanding the embarrassment caused by Tay, the racist chatbot, believes that chatbots will replace apps. He is not alone in heralding a new chapter in digital engagement. As natural language understanding and processing come of age, the potential to deliver human-like conversations between humans and machines is clearly suggesting some amazing opportunities.
Facebook Messenger has been quick at the game, by providing a platform and tools for developers to build chatbots for its service. There are currently more than 11,000 chatbots waiting to pick up a conversation with you on Messenger. With more than 900 million users Messenger is a great medium to reach potential customers and build relations. So the temptation to go all chatbot is great. Especially, as a plateau seems to have been reached in the appetite of consumers to download “one more app”. So if that’s the opportunity, what are the challenges?
I was recently talking at a company event, about the potential and future of AI, when I came upon the subject of chatbots. We take for granted that we can only communicate with computers via screens, keyboards and mouses, clicking or tapping. This kind of user interface has been with us since the dawn of digital computers in the 1940s, evolving with the times. I am old enough to remember the excitement that Macintosh stirred in every MS-DOS user (anyone remember MS-DOS command lines?) with its beautiful interface, and elegant mouse. At last, here was a machine with which it was easy to interact and communicate with! Touch screens on smartphones and tablets delivered a new paradigm, and a new kind of dialogue. Less typing, more tapping. Megatons of grey matter power and creative talent have been spent – and are still being spent – on forging great user interfaces that are intuitive and easy to use. But all that effort has been trying to solve for a single problem: that computers are not human. They speak code, we speak natural language.
But no more. The time when we can have a natural language conversation with our smartphone, tablet, speaker, fridge, car – you name it – is afoot. In a world where machines can understand what we are saying and can respond in kind, there is no need to do much in the way of user interface design. Like jumping in a time machine, we are transported back to the age of simple telephones. No need for managing apps. Just dial a chatbot and speak clearly so it (he/she) can hear you well.
But as I was going through my self-induced exhilaration about how bright the future of human-machine relationship looks, I was interrupted by a member of the audience. “I don’t really like having a conversation with a machine”, she said. “I prefer speaking to a human”. The audience applauded. They, everyone, has been through the painful frustration of automated call centres. Bring back the humans! No more automated responses! Please!
The lady had a point. In the end, it will all be about the user – read “human” - experience. The Messenger chatbots are not yet there when it comes to delivering great experiences, but they are evolving. The more humans use them, the better chatbots will become. Machine learning, as the terms suggest, is a curve not a point. It takes time, and data, lots of data. But once the chatbots pass the Turing test and become indistinguishable from a human interlocutor, would the lady in the audience, and others like her, accept having a conversation with them?
A neo-luddite viewpoint would be that there is something inherent in human psychology that will ultimately set a limit to how far we will accept AI in our lives, chatbots including. We are already seeing this in our reaction to human-like androids, a phenomenon called Uncanny Valley. Perhaps there are cultural factors at play a role too. Western cultures tend to be more suspicious and fearful of machine intelligence, compared to oriental ones. Or the barrier to acceptance could be generational. What we consider “normal” today is constantly shifting. The “new normal” in a few years could be having simultaneous conversations with artificial and non-artificial beings without feeling, or caring, for the difference; a paradigm that Facebook is following by introducing chatbots amongst our friends in Messenger. Time will tell…