ORLANDO, Fla. -- Artificial intelligence chatbot vendors may hype machine learning tools to enhance customer service, but call center leaders aren't necessarily ready to trust them in the real world.
Part of the reason is call centers are judged by hard-to-achieve performance metrics based on volume, efficiency and customer satisfaction. Once a call center performs successfully against those expectations set by management, it's hard to convince leaders to entrust call center chatbots with the hard-fought, quality customer relations programs they've built with humans.
"I don't anticipate them having any kind of utility here," said Jason Baker, senior vice president of operations for Entertainment Benefits Group (EBG), which manages discount tickets and other promotions for 61 million employees at 40,000 client companies. Baker oversees EBG customer service spanning multiple call centers.
"We strive for creating personalized and memorable experiences," Baker said. "A chatbot -- I understand the reason behind it, and, depending upon the type of environment, it might make sense -- but in the travel and entertainment industry, you have to have the personalized touch with all interactions."
Artificial intelligence chatbots were the most-talked-about technology at the ICMI Contact Center Expo, with a mix of trepidation and interest.
Navy Federal Credit Union employs "a few bots" for fielding very basic customer questions, such as balance inquiries, said Georgia Adams, social care supervisor at credit union, based in Vienna, Va.
Her active social media team publishes tens of thousands of posts and comments annually on Twitter, Instagram and Facebook without the help of artificial intelligence chatbots, but "they're on the horizon." She stressed that call center chatbots must be transparent -- identifying themselves as a bot -- and be empowered to transfer customers to human customer service agents quickly to be effective.
"It's coming, whether you want it or not," Adams said. "We're strategizing [and] looking at it. I certainly think they have a lot of value, especially when it comes to things that are basically self-service ... but if I'm talking to a bot, I want to know I'm talking to a bot."
Report confirms fleeting nature of chatbots
Gartner researchers predicted that in the next four years, many companies will try chatbots -- and dump them. Among the predicted trends:
- Seventy percent of customer interactions in 2022 will involve machine learning applications, chatbots or mobile messaging -- almost five times more than 15% in 2018.
- Twenty percent of customer service interactions will be completely handled by AI in 2022 -- quadrupling 2018 levels.
- Phone-based communication will drop to about 10% of customer service interactions by 2022, but human agents will still be involved more than 40% of the time.
But it's not all sunshine and roses: By 2020, Gartner predicted that 40% of bot and virtual assistant applications launched in 2018 will have been abandoned.
Source: Magic Quadrant for the CRM Customer Engagement Center (May 2018)
Call center chatbots not gunning for humans' jobs -- yet
Another part of the reason call center personnel might be wary of chatbots -- true or not, fair or unfair -- is robotic automation will eventually take the humans' jobs. This idea was dismissed by neutral industry experts such as ICMI founding partner Brad Cleveland, who said alternative customer service channels such as interactive voice response (IVR), email, social media and live chat each caused similar panic in the call center world when they were new. But none of them significantly affected call volumes.
"We hear predictions that artificial intelligence will replace all the jobs out there," Cleveland said, not just in customer service. "If it does, we're definitely going to be the last ones standing in customer service. But I don't think it's going to happen that way at all."
Cleveland said he believes artificial intelligence chatbots will likely have utility in the near future, as technology advances and call centers find appropriate uses for them. Machine learning tools that aren't chatbots, too, will make a difference, he said.
One example on display was an AI tool that can be trained to find -- and adapt on the fly -- pre-worded answers to common, or complex and time-consuming, customer queries that a human agent can paste into a chat window after a quick edit for sense and perhaps personalization. The idea is they get smarter and more on point over months of use.
But even live chat channels have their limits when they're run by humans, let alone artificial intelligence chatbots. Frankie Littleford, vice president of customer support at JetBlue, based in Long Island City, N.Y., said during a breakout session here that her agents have to develop a sixth sense about when to stop typing and pick up the phone.
"You know in your gut when to take it out of email or whatever channel that isn't person-to-person," Littleford said. "You just continue to make someone angrier when you're going back and forth -- and let's face it, a lot of people are really brave when they're not face-to-face or on the phone ... If your agents are skilled to speak with those customers, you can allow them to climb their mountain of anger and then de-escalate."
Vendors hold out hope
ICMI attendees weren't fully buying into the promise of AI chatbots, but undeterred software vendors kept up the full-court press, attempting to sell the benefits of automation and allay fears that chatbots will eventually replace attendees' jobs.
"We don't use [AI] to replace human work," said Mark Bloom, Salesforce Service Cloud senior director of strategy and operations, during his keynote, adding that organizations that attempt to replace people with AI tools haven't been successful. "We want to augment the work our people are doing and make them more intelligent. That is how we are moving forward."
Kaye Chapmancontent and client training manager, Comm100
Setting up call center chatbots will require extensive training in test environments -- just like human agents do. Once they're trained, they require maintenance and updating, but they will solve another vexing problem for call center managers -- employee turnover, said Kaye Chapman, content and client training manager for chatbot vendor Comm100, based in Vancouver, B.C.
"You could train a new employee, and they could leave tomorrow," Chapman said. "A bot is not going to give up and leave, it's not going to get sick, and it's so scalable."
Bob Furniss, vice president at Bluewolf, an IBM subsidiary known for Salesforce automation integrations that's based in New York, said he believes artificial intelligence chatbots are coming, and AI in general will change both our personal and work lives. He said the potential is there for AI to help ease call center agents' workload -- up to 30% of the simplest customer queries -- similar to the promises of IVR and the other channels when they came online in the industry.
Just like all other call center systems, Furniss warned that anything AI-powered will require attention and maintenance to attenuate its actions and keep abreast of changing workflow and updated customer relations strategies.
"This is just like any other technology we have in the contact center," Furniss said. "You don't set it and leave it, just like workforce management [applications]. There's an art and a skill to it."
Hospitality vendors are eyeing AI to make things smoother for travelers