Bots are hot. Or, as Microsoft CEO Satya Nadella says, they’re “the new apps.”
But what caused this sudden rise to fame? Chatbots aren’t exactly new: Their digital origins lie in text-based video games such as Colossal Cave Adventure and the Ultima series, which stoked the imaginations of countless computer enthusiasts in the 1980s. But, while graphical user interface technology killed that particular trend, chatbots have never really gone away, existing as “chatterbots” in services like SmarterChild and Spleak, which entertained users in the early ’00s on instant messaging clients such as AIM. These ancient pieces of software were the predecessors of today’s Siri and Google Assistant, and have now moved into the spotlight once again as chatbot fever sweeps Silicon Valley.
One person pushing the boundaries of modern chatbot technology is Daniel Ilkovich. Ilkovich founded Dexter in 2015, with the aim of creating a platform that makes building chatbots possible even for people without technical skills. “Dexter is trying to bring the tools required to build chatbots to their rightful audience: writers,” he said at the Rakuten Technology Conference 2016 in Tokyo. “Writers are the new designers.”
But Ilkovich believes that the significance of chatbots is far more profound: “Take a small business, for example, that gets a lot of interactions on various social networks. How do you help that business automate some of those interactions?”
The answer, Ilkovich says, is a chatbot that is programmed to talk to customers within social networks. A consumer might send an enquiry to the small business’ Viber or Facebook account, for example, and receive a response generated by a chatbot, instantly.
Still, building a chatbot generally requires more resources than are available to your average small business: “a huge engineering team,” as Ilkovich says. The aim of Dexter, which Rakuten Ventures invested in earlier this year, is to make chatbots easier to build, he explained, and hence more accessible.
One way that might be possible is by using templates. Business owners would be able to use pre-made chatbot templates that would eliminate the need to build entire conversation trees from scratch. A restaurant template, for example, would have most interactions pre-built, with a business owner needing only to fill in details specific to their business, such as menus, locations or parking availability, before deploying.
Still, Ilkovich stresses the need for some ground rules when dealing with chatbots. The first is to keep expectations in check: “People who try to make bots are typically trying to get them to act like a human,” he says, but that’s just not possible right now.
What chatbots can do, he explains, is predict appropriate responses to questions – and they are getting better at it every day. “While you can’t teach a computer to truly understand a human, if you teach it enough situations where the human responded correctly, you can predict what an appropriate response would be with increasing certainty,” he explains.
Nevertheless, he continues, it’s important that users be informed when they are talking to a chatbot and be given the choice to opt out.
If they are done well, Ilkovich predicts, chatbots will be around for a long time.
Real-time text-based chat is fast becoming humanity’s preferred method of long-distance communication, he says. “We as humans no longer like the burden of voice. No-one calls each other anymore!”