Chatbots Rising

Chatbots have actually been around in various forms for a long time.  For those of a certain age (those who remember dying of dysentery while playing Oregon Trail, for example) there was Clippy, the cute and annoying assistant that would pop up in versions of Microsoft Word.  Clippy was trying to be helpful, but he lived alone in a world in which users were not yet used to dealing with bots and clamored always to talk to “a real person” (preferably in their own time zone and with their accent, please).  Time has marched on and Clippy is no longer with us, but his improved descendants are, and more and more, we aren’t refusing to talk to them (though we still seem to be rude to non-native speakers who take our calls).  Before too long, we might not be able to tell the difference between them and “real people.”

Why chatbots?

The chatbots that exist now are using machine learning and algorithms to constantly improve their performance.  The reality is that so many customer interactions are not complex exchanges requiring thoughtful human discourse and cogitation, but simply exchanges of data, with the chatbot serving as the cipher.  Jetstar, a budget airline operating mostly out of Asia, has a chatbot called “Ask Jess” who can confirm your flight times or help you make changes to your existing itinerary, functions in the past that were restricted to human operators, who imposed a bandwidth restriction on the company (we haven’t yet evolved as a species to take multiple calls simultaneously).  “Jess” can handle hundreds, if not thousands of inquiries simultaneously, satisfying customers who simply have some basic inquiries while creating well-worn, traceable paths to the caches of big data that firms more and more have but less and less know how to use effectively.

Where will they live?

While numbers vary, it seems that conservative estimates put 25% of the world’s population on mobile messaging apps like Whatsapp, Facebook Messenger, and WeChat, while others put that estimate north of 60%.  Regardless of which estimate you use, it’s clear that these 1:1 messaging platforms are more and more where internet users are spending their time, rivaling time spent on social networks.  The bots that currently live on Facebook Messenger, which can allow you to order flowers, check the weather, or order an uber, for example, are simply imitating functions that have long lived on Asia’s behemoth WeChat, which even has its own payment infrastructure.App fatigue means that users may not even remember if there’s a native app installed on their device to interact with a certain company or brand, whereas a chatbot can easily be called up and queried within a messaging app that users access dozens, if not hundreds of times per day.

How will they be built?

As we’ve seen over and over in the internet age, there’s a gold rush going on.  It happened in the early 2000s with domain names and it happened in the not too distant past with apps.  People see opportunity and get excited and rush to buy their digital pickaxes and sifting pans and many useless websites and irrelevant apps were created.  So it will be with chatbots too.  This is even more so the case because the cost to develop a chatbot is much less than to develop an app, and the plug-in functionality is even simpler.

What comes next?

All this machine learning and AI points to a future in which humans may not be able to tell they are chatting with a bot and these technologies will be freed to pursue more than basic frontline customer interactions, but start to provide cognitive assistance currently only provided by a competent class of humans.  One such scenario?  Imagine telling a chatbot to keep a lookout for flights less than $500USD going to Dublin from Los Angeles from March-June, and to buy a two-week roundtrip ticket anywhere in that time horizon, using your credit card information, if such a fare occurs.  The technology exists to do that now – but not in the frictionless form of a chatbot.  Not yet, anyway.