Sales/Customer Service Chatbots
An Emerging Technology Worthy of Close Monitoring
Even though IBM’s weather chatbot is powered by its famed artificial-intelligence computer, Watson, the bot is intended to operate using only a limited number of conversational phrases.
While chatbots—computer programs that conduct conversations via audio or text—have captured the imaginations of businesses hungry to automate conversations on the web, it will still be a while before any “digital friend” inside your computer can truly mimic human conversation. Even so, businesses which ignore chatbots do so at their peril, given that there is a great deal of money betting chatbots will emerge, sooner or later, as conversational charmers.
Why the Wait?
So far, the digerati remain flummoxed about how to enable a machine to understand all of the nuances and subtleties inherent in everyday communication. Probably the most famous of recent failures was Microsoft’s Tay, a chatbot that the company added to Twitter last year as an experiment in chatbot/human interaction.
Microsoft’s goals were lofty. It backed Tay with some heavy-duty artificial intelligence (AI) software designed to enable the chatbot to improve its conversational skills over time by randomly chatting with anyone who stopped by on Twitter and “learning” from those interactions. Unfortunately, many of the people who first interacted with Tay thought it would be funny to use text conversation to transform it into a rube. Within 24 hours, Tay was spouting racist and misogynistic utterances all over Twitter—with gusto.
“Microsoft’s latest attempt did not end well,” said Ekim Nazim Kay, founding partner of Botego (www.botego.com), a maker of chatbots. “Our conversations are hectic, run-on sentiments, and there is more to what is being said than the words that are vocalized. Pitch, emphasis, colloquial variations—understanding and interpreting what is actually being said takes more than just parsing out the words.”
Vexing, for sure.
But Kay, along with an army of other chatbot creators, is convinced that it’s only a matter of time before we won’t be able to imagine how we lived without chatbots, and that the challenge to creating a bot that can truly engage in natural conversation will be realized.
Kay promised that, once developers figure out the methods, we will see an explosion of bots that will be so well versed in human conversation, you won’t be able to distinguish between a bot and the real thing. Between here and there, forward-looking businesses will still want to follow the chat space very closely, given that major players in technology are making significant bets that chatbots will indeed flourish.
Key among those is Facebook. It lit up the chatbot world last year by introducing its Facebook Messenger Chatbots Platform (https://developers.facebook.com/blog/post/2016/04/12/bots-for-messenger), which developers can use to build chatbots to interact with consumers directly in Facebook Messenger.
The move was especially significant, given that Facebook Messenger’s growth is on a tear, as are many messaging services such as Kik, Line, WhatsApp, WeChat and iMessage. Indeed, according to a 2015 report released by BI Intelligence
(www.businessinsider.com/the-messaging-app-report-2015-11), more people are now using the top four messaging apps every month than the top four social networks.
Essentially, messaging services—once simple services for exchanges of messages, pictures, videos and GIFS—have evolved into expansive ecosystems, complete with developers who are determined to build out fully developed digital communities around the messaging. In a phrase, all those chatting smartphone and computer users on Facebook Messenger and similar services represent an incredible opportunity for businesses, according to Kay.
In fact, with Facebook alone, businesses now ideally have access to more than one billion active users each month via chatbot. No wonder Facebook’s decision to add a chatbot platform to Messenger—and create a common, global, digital neighborhood where consumers and commercial chatbots can easily interact with one another—is still sending reverberations throughout the chatbot world.
A commercial entity now has the capability to interact directly on Facebook Messenger, Kay said. Before, the same entity had to wait for a user to come over to its website and click a link or a button to engage in chat with that user.
Many businesses aware of Facebook’s new chatbot platform have responded with keen interest. But they’ve also done so prudently, deliberately setting expectations low regarding what their chatbots can and cannot do. Essentially, they’re eager to avoid creating another Microsoft Tay. Domino’s, for example, offers a chatbot on Facebook Messenger that users can call up to order a pizza on the fly, without ever having to leave the Facebook Messenger universe.
“We’re delighted to allow our customers to order directly from Messenger,” said Nick Dutch, head of digital for Domino’s. “With one billion monthly active people using Messenger each month, it’s the obvious choice for Domino’s.”
|Chatbox Info Resources
But there’s a catch: To use Domino’s chatbot, you need to visit the Domino’s website prior to any chatting and pre-select the kind of pizza you like. Your pizza, and the fixin’s you like, are actually stored in your profile on Domino’s website well before you ever interact with a chatbot.
Sure, you can ultimately interact in a basic way with the Domino’s chatbot on Facebook Messenger when your profile has been established and you’re hungry for pizza. But the chatbot already knows what kind of pizza you want, based on the pre-stored profile. So in reality, Domino’s chabot is basically a web form in disguise.
Meanwhile, another successful chatbot taking a similarly understated approach is IBM’s weatherbot on Facebook Messenger (www.theweathercompany.com/newsroom/2016/10/25/weather-channel-launches-bot-facebook-messenger-powered-ibm-watson). Users can interact with the chatbot to retrieve info such as current weather conditions, forecasts, severe weather notices and the like. But even though the chatbot is powered by the artificial intelligence of IBM’s famed Watson supercomputer, actual conversations with the bot are limited to a small number of phrases.
Still another early, conservative player in the chatbot space is Mastercard. It experimented with the technology in 2016 and is promising to add chatbots to messaging services such as Facebook Messenger sometime this year.
“At Mastercard, we believe that AI-driven conversations between companies and their customers can drive better customer experiences in places and platforms where consumers are already engaging,” said Kiki Del Valle, a senior vice president at Mastercard.
Sounds wonderful. But like IBM and Domino’s, Mastercard’s chatbot ambitions are, in reality, very modest. The company is promising only that its chatbots will enable customers to get answers to basic questions about their accounts, such as details about their purchase history, spending limit and cardholder benefits.
Bottom line: Without question, the gee-whiz factor on the aforementioned chatbots is a bit lackluster, especially given the conversational prowess of the fictional robots that loom large in the collective mind—robots such as “Hal,” the evil supercomputer from 2001, A Space Odyssey, and “Samantha,” the supercomputer that Joaquin Phonenix’s character fell in love with in the more recent flick, Her.
But even these chatbots are a healthy start, Kay insisted.
Businesses that get hip to what’s happening now with chatbots will be in a much better place down the line, when chatbots break though with the capability of truly robust conversation and everyone else is playing catchup about what the heck is going on.