The idea of automated customer service is nothing new – from speech recognition technology on dreaded call center phone lines, to Apple’s groundbreaking Siri personal assistant system, AI has been in our lives and in our pockets for a while now. Yet, young as 2016 still is, it’s already defined itself as the year of the Chatbot – from Microsoft’s experimental and indeed controversial (more on that later) attempt to engage millennials with the creation of ‘Tay’, a chatbot designed to imitate a teenage girl, in March, to April’s recent inundation of announcements from companies eager to trial Chatbots, this year could be the year Chatbots go mainstream.
But in a world where the search for authenticity seems increasingly futile and more and more of our meaningful interactions take place via a screen, will the public really welcome Bots? Are they even ready for widespread release? And how are businesses going to capitalize on them? Read on for Tej Kohli Blog’s analysis.
What is a ChatBot?
Put simply, a Bot is a piece of software which performs automated tasks and simple, time-consuming or repetitive errands, and this is exactly what a ChatBot does – except (you guessed it), it simulates human interaction. One of the earliest and most popular examples of a Chatbot is CleverBot – an algorithm set up almost 20 years ago to have conversations with humans. It’s not programmed with responses, but rather learns from human input. I found its responses to be mixed (I asked CleverBot what it could do for my business, and it responded with a demand that I make it a cup of tea, perhaps demonstrating AI’s eventual intent to overthrow their human masters), but in a formal Turing test in 2011, CleverBot’s responses were rated 59.3% human, while real human’s responses were rated 63.3% human. Considering the Turin test was developed to answer the philosophical question ‘Can machines think?’ that’s scary stuff.
It’s a question Tej Kohli Ventures has been asking for some time now – does the advancement in AI technology mean that Bots are ready to go to work for corporations? There’s several different ways in which Chatbots are being trialed to increase productivity – China has successfully used WeChat for a while now to enable customers to complete basic tasks such as ordering food, paying restaurant bills and transferring money and it looks like Western companies are hoping their bots will mimic these functions. Absa, Barclay’s South African subsidiary has announced their intention to trial Chatbots for customer service queries and Shopify recently acquired Kit, an all-round marketing masterpiece that can do everything from email customers to help you set discounts and handle 404 errors. Facebook recently announced that their new messenger platform is open to Chatbots, which could be a marketing game-changer considering that messengering apps now significantly outperform social media in terms of engagement. With this seminal launch, it’s looking likely that we’ll see a surge in Chatbot-based marketing – but are Chatbots really what consumers want, and will they add value to your company?
There's no doubt that, for large corporations, if Chatbots could replace customer service systems or traditional marketing, they could save untold millions. But only if they can add value to the user experience rather than repelling users, and this is where AI can fail in the most unexpected ways.
I mentioned Microsoft’s recent attempt at engaging younger customers via Twitter Chatbot in March – while this experiment proved highly successful in terms of engagement, it can also be seen as a cautionary tale. ‘Tay’, Microsoft’s Chatbot character, used a mixture of AI, pre-programmed responses and publically available online data to formulate responses – but, like Cleverbot, she also learned from human engagement. It took just two tweets from an online troll for Tay to begin spewing antisemitism, and the project was shut down after 16 hours following a barrage of racist and extremist tweets from the young AI. This is a very real fear for any company planning on programming their Bot in this fashion – while Chatbots may be a novel way of engaging with a consumer, part of their novelty value is in their potential for misuse. It is also likely that the novelty value which provokes customers to engage with Chatbots will wear thin very quickly as the technology becomes widespread, meaning its marketing value could be short-lived.
There’s no doubt that ChatBots can add value to a company by performing menial, repetitive tasks, and China’s widespread success using the technology indicates it could be equally successful elsewhere. But it is worth bearing in mind that China favours more traditional, multiple-choice methods of helping customers to complete tasks rather than simulating human conversations – and, from the limited evidence available thus far, it seems likely that it is this form of ChatBot that will prevail.