Over the last month, companies from Microsoft to Facebook began rolling out chatbots, handy tools meant to make life easier in a variety of ways.
For those who remember Clippy—the intelligent user interface introduced by Microsoft in 1997 to assist users with Office by way of an interactive animated character—the news about new chatbots may have been met with an eye roll. Clippy drew strong negative responses from many users and was eventually, slowly phased out. But technology, and AI in particular, has come a long way since Clippy, and the time may now finally be ripe for bot-driven assistants and services.
In early April, messaging app Kik launched a chatbot store, Kik Bot Shop, that will allow users to download app-like bots that directly connect them to brands. Shortly after Kik's release, Facebook introduced chatbot services to its Messenger app that can automate customer service tasks currently offered via phone and e-mail. The Messenger bots can provide customized communications from businesses, such as receipts, shipping notifications and live automated messages "all by interacting directly with the people who want to get them," according to Facebook.
Meanwhile, Microsoft introduced a new bot called Tay. And although Microsoft quickly ran into some issues with Tay (the chatbot was trained by Internet trolls to respond with racially insensitive language and misogyny) it doesn't change the larger fact that Microsoft and others believe the world will soon move away from apps into a phase dominated by bots.
"I think we all agree that the future cannot be more apps on your phone," said Dennis R. Mortensen, CEO and founder of x.ai, a New York-based artificial intelligence company. "Going from 120 apps on average to 240 apps isn't progress."
X.ai is focusing on intelligent agents, as they call them, though Mortensen is quick to point out that names are not important.
"We can talk about whether we call them AI, intelligent agents, bots, what have you," he said. "We can probably cluster them all together and say simply that they represent that next paradigm."
The new paradigm is about relaying, preferably in natural language, what a user needs and having technology take care of it with the least amount of clicks possible. The best way to accomplish this? Many think its bots.
Bots at work
What will the "rise of the bots" mean for the enterprise? At first, companies are most likely to apply them to routine customer service experiences. In many cases, they already are.
"When you interact with an organization's live chat, much of the interaction is automated with only a small portion controlled by a customer service agent," said Taylor M. Wells, Ph.D., assistant professor of MIS in the College of Business Administration at California State University, Sacramento.
But, potentially, bots could mean the introduction of new forms of automation that lead to cost savings to innovative ways of engaging with customers and products. Or bots could allow for new approaches to the overall "customer experience."
Bots "represent a new path to reach consumers, making it easier to integrate their preferences with revenue opportunities, for example," said Sanjay Mehta, CMO of ThousandEyes.
Beyond the customer experience, bots can also make some of the tasks businesses perform everyday easier and more efficient. For example, on average, it takes 2.2 days for humans to set up a meeting, according to Mortensen.
"I've been out of college for 20 years, but the way I set-up meetings 20 years ago is exactly how I set-up meetings today," said Mortensen. "It's like e-mail ping pong. The only thing that really changed was the email client and that didn't even change much."
To try and solve the problem, x.ai developed Amy and her male counterpart, Andrew, virtual assistants that schedule meetings. X.ai's robots have reduced the average time it takes to set up a meeting from 2.2 days to 1.8 days, and that delay is still mostly due to the humans, said Mortensen.
X.ai plans to make its intelligent assistants available to the public later this year, and they'll eventually be able to handle much more than just meetings. For example, if a business traveler needs to figure out the cheapest, earliest, most convenient flight to Miami from JFK, they should conceivably be able to relay that request to a virtual assistant.
"What I don't need to do is get on my laptop, open it up, authenticate, find Expedia, open the app, type in JFK, type in the destination, see a list of query results, pick one, etc.," said Mortensen.
In other words, technology can handle much more than we give it credit for on its own. Now it's just a matter of connecting what someone wants or needs with how to deliver it.
"I want the job done and I don't want to do it. It could be something as simple and easy as a phone call," said Mortensen. "What I don't want is to email back and forth for a day and a half to figure out that today at 5 o'clock we talk can talk on this number. That is just a waste of my time. So instead of hiring someone and paying him or her $50,000 a year to manage my calendar, I use a bot to handle that for me."
And it's not just about the cost savings, added Mortensen.
"Shuffling that duty from one live person to another does not remove the pain. It just moves the pain. It's saying that somebody is supposedly is more valuable than somebody else," said Mortensen.
"There are many businesses where that's exactly what you do, whether that is cleaning your apartment or scheduling meetings or cooking your dinner at a restaurant," said Mortensen. "We think there is an opening today to once and for all democratize a particular human task, such as setting up meetings, to the extent where it completely disappears."
Challenges ahead
That's not to say it will be clear sailing for chat bots moving forward. As we learned from the Clippy debacle, if a company is going to do it, they need to do it right. Already, the latest bots have been followed by a wave of criticisms about "bot latency," according to Mehta.
"Some reports (have cited) half an hour wait times, instead of the expected five seconds," said Mehta. "The good news is that bot makers can get ahead of infrastructure challenges that disrupt the online experience. For example, it's possible for them to understand every network dependency that impacts bot latency. Knowing and optimizing this means they can focus on building the best possible product."
"The challenge from an organizational perspective is whether and how to have these bots interact with the public," said Wells. "Obviously there are strong dangers to this (as Microsoft found out with Tay). We are in the area where the handler of the corporate Twitter account can singlehandedly build or harm a brand's credibility. Will organizations be willing to trust these bots to speak for them?"
According to Mortensen, accuracy will be critical to the survival and success of bots.
"There are many activities where you don't need high accuracy, like conducting a Google search," he said. "If the results aren't what you expected, you just tune the query a little bit because you expect that somewhere on the Internet there is something which you're looking for."
But with bots, accuracy is absolutely critical.
"If I tell you that we can only schedule eight out of 10 meetings perfectly, then you probably can't use [our virtual assistant] Amy, because what use is it if it doesn't work all the time?" asked Mortensen. "It's like bringing a self-driving car to market with a footnote that states 'For every 10,000 miles we're going to hit a pedestrian.' It just doesn't work."