Setting the Right Expectations for Chatbots

Armies of “digital workers” are poised to take over the front lines with customers, but they need rigorously organized data to perform well, and often a helping hand, say experts at Earley Information Science roundtable

The numbers are very real even if the workers are virtual. Spending on Enterprise Intelligent Assistants, aka chatbots, is now exceeding $1 billion a year and is expected to grow to $4.5 billion by 2021, according to Opus Research. And a recent survey of 100 information management professionals found that 51% are getting ready to put chatbots to work with customers. 

But it is still early days in the development of artificial intelligence (AI), and companies have to be realistic about the kind of work that chatbots can, and can’t, handle, and the support they need to be effective, according to a panel of AI and knowledge management experts at an Executive Roundtable discussion hosted on March 29 by Earley Information Science Corp. (EIS), a leading consulting firm focused on organizing information for business impact.

The data that is the chatbot’s lifeblood has to be structured in a way that is fully compatible with a company’s business, humans still have to help out, and tasks that involve a lot of context and nuance may be out of a chatbot’s reach, at least for now. That said, digital workers are already earning their keep, and are taking on more and more responsibilities as they get “smarter,” the panelists said, adding that companies would do well to go on a hiring spree so long as their expectations are properly set.    

Chatbots “are evolving very very quickly,” said Seth Earley, founder and Chief Executive Officer of EIS. “But right now they are not necessarily easy to use. There is no magic – chatbots require human intervention and knowledge engineering. They are an access mechanism to get to information,” which has to be carefully collected and organized so that the bots can understand and accomplish their tasks and learn new ones.

The roundtable discussion, “Digital Workers, Chatbots and Customer Service: Reality Versus Aspiration,” was led by David Hatch, Senior Vice President for Marketing at EIS. Besides Seth Earley, the panel included Tom Davenport, a professor of information technology and management at Babson College and author of “Only Humans Need Apply: Winners and Losers in the Age of Smart Machines”; Dan Miller, Lead Analyst and Founder of Opus Research; and Mark Beccue, Principal Analyst at Tractica.

The discussion was the first of a four-part EIS series, “The Age of the Chatbot Is Coming: Are You Ready?” with new roundtables set for April 12, April 26 and May 10 (details below). EIS held a related roundtable in January, “Virtual Assistants and Chatbots: The Next Big Thing in Customer Experience.” 

At this stage of bot development, companies should target “low-hanging fruit” to start out, Tom Davenport suggested, noting that so-called transformational projects involving bots that are pushed down from the top have a low success rate, particularly if the company is the first in its industry to make such a big move.

The shift to digital workers “may well have a devastating impact on the job market in the future,” but right now, Davenport said, there is little job loss and no pure automation. “It’s a hybrid machine/human process.” The highest return on investment (ROI) is coming from the most prosaic applications and technologies, such as robotic process automation (RPA), and not from layoffs.

As companies bring in bots, they should find the speediest ways to produce some business results while keeping an eye on costs, said Opus Research’s Dan Miller. Think about what works out of the box, but also factor in the ongoing expenses for professional services to collect, organize and reshape the data. Try to leverage existing resources, including key data, operational personnel and subject matter experts. And, of course, figure out the right tools for development, analytics and reporting, including metrics to track success and failure.

“Some fairly complex implementations are already out there,” Miller said. In some cases, bots “can do more than just low-hanging fruit.”

What are the “compelling market drivers” that are opening the market to chatbots?

Tractica’s Mark Beccue noted consumer desire for 24/7 support and self-service and an aversion to live interactions. The use cases are also expanding – into financial services, for example, with the onboarding of new customers and automated wealth advisors, and into healthcare, with coaching services for patients with long-term illness.

Chatbots will not be taking over the world any time soon, Beccue said. Advances are being made in understanding language, “but AI will continue to be challenged by the context of human speech.” (Consider this statement: “I would kill for a cookie.”)

For now, at least, the trend is toward seamless handoffs from bots to live agents when the task, or the language, becomes too difficult for the bot to handle – and back to the bot when it makes sense.   

The goal, in these early days of the era of the smart machines, Beccue said, is for “chatbots to live in harmony with live agents.”

The roundtable featured a real-time survey of the webinar attendees:  

  • Nearly a fifth of the respondents, or 18%, said that bots are a major priority (supported by funding and executive buy-in) and 37% said they are important but still require a business case. Almost half, or 45%, said bots are not a priority at this time.
  • None of the attendees has been using bots for more than a year, but 6% have recently launched their first wave to deal with customers, 27% have begun testing in one area of the business, another 10% are testing the business case and more than half, or 54%, are starting to investigate how they can use bots.
  • When asked where chatbots are now being deployed at their companies, 36% cited customer/member service, making it the leading application. IT help desk/tech support accounted for 15% and selling in digital commerce 13%.

Please use these links to access the roundtable and EIS resources about chatbots and AI.

Finally, here are details for the upcoming one-hour roundtables in the chatbot series, all live at 1 p.m. (Eastern) and recorded for later access:

April 12 – The Role of Intelligent Virtual Assistants in Healthcare: How virtual assistants can help solve the “adherence gap” and more

May 10 – How Financial Services Organizations Are Using Chatbots in Customer Service: Going beyond basic banking transactions to meet the needs of new audiences

June 7 – Conversational Commerce and Virtual Sales Assistants: How the age of the virtual sales agent is changing the face of commerce

To register for any of these webinars, contact Sharon Foley, at sharon.foley@earley.com

 

 

Leave a Reply

Your email address will not be published. Required fields are marked *