It may be hard to believe all the hype out there about chatbots – if only because there’s just so darn much of it out there right now to believe. But I’m here to attest that bots are not just a fad. Based on the early returns we’ve witnessed since developing one to use with our community of sports fans, I think they truly represent a promising, transformative new interface method ready to enable users to get more out of their technology.
Of course, not all applications are equally appropriate candidates for being driven by a chatbot interface. There are apps in our lives nowadays that have reduced ordering a pizza or summoning a ridesharing service to a just about a single click – interfaces that have purposefully eliminated the necessity of having a conversation. A simple rule of thumb for bot technology is that it should be applied where it allows users to communicate more efficiently, enjoyably, and easily.
Chatbots have a tremendous ability to reduce user effort – and error – in the completion of tedious tasks. For example, when a user is required to fill out a complicated form where data must be entered in the correct fields and correctly formatted, it’s much more efficient and pleasant for them answer a bot’s questions and let it handle the careful data entry than to grapple with the form directly. Interfaces for completing medical, insurance or financial forms, making hotel room bookings, and other such tasks can be greatly improved by introducing bots to do the heavy lifting.
They’re also vastly more efficient in giving a user quick access to information that has become siloed within multiple apps or locations.
For example, a user who wants to causally and rapidly check sports scores, the weather, or their day’s schedule can get this information with three questions rather than opening and closing three apps – and with voice controls, the person can do all this while putting their shoes on in the morning. Chatbots are also superior for completing single complex tasks that require cross-referencing information, and can reach useful conclusions on the user’s behalf. A traveling family might ask a bot interface to “Find a child-friendly restaurant along our current route,” and the bot will gather the needed data and make a satisfactory selection, possibly even incorporating restaurant ratings and offering to provide directions to the destination.
But just as not all applications are not ideal for bot additions, not all chatbots are created equal. The bot interfaces that best succeed will likely be those designed with the most careful eye on the quality of the user experience. This will be driven in large part by the naturalness of the bot as a conversationalist. Natural human speech is a messy business – people speak in slang, use pronouns, and change topics at a whim. However, if a chatbot needs to give answers along the lines of “I didn’t catch that” too often, users will give up on it.
Bots capable of robust natural language understanding (NLU), that can readily handle the subtleties of human speech, are, in my opinion, an absolute necessity for delivering an optimal user experience. For our own bot creation, we’ve used the API.AI platform to power NLU-enabled chatbots, because it is able to make sense of the colloquial language, references, and rapid conversation-switching that users require solutions for.
In the bigger picture, the rise of bots does represent something of a pendulum swing back towards users employing their voice and conversational skills to achieve tasks rather than their fingers. In a world where it’s now rare for individuals to engage in a phone call when a text will do, chatbots fully capable of convincingly-human conversation are demonstrating the power of language as an extraordinarily effective computer interface.