Action Week 2018: Can CSPs move AI from ‘sci-fi’ to deployment?

When customer experience or operations teams within communications service providers (CSPs) want to adopt artificial intelligence (AI), they first must find a way to ‘sell’ it to finance executives, and perhaps more importantly to the employees who could be displaced by the technology. Neither is an easy job, according to a panel of experts gathered here at Action Week in Dallas.

Jerrid Hamann, Digital Customer Experience Strategist, Verizon, who has worked for the company for about a year, spoke about his experience at another telco where he was trying to implement AI for customer service. It was a struggle.

“I was trying to convince finance that we needed the new tools to improve customer experience,” he said. “Either they were very skeptical and suspicious saying it sounds like science fiction and is not something we want to invest our money in…or at the other end of the spectrum they say, ‘Oh wow, we can save that much money? Let’s lay off the entire contact center’. When you get that kind of reaction you have to dial it back and explain that it’s something that has to be phased in.”

Calming fears

The sell was even more difficult with the customer service teams that could be displaced by AI chatbots.

“When we tried to introduce the idea, we certainly got a lot of pushback from the folks in the contact center who said, ‘This looks like a threat to our livelihood’,” Hamann said. “We tried to find ways to soften it, telling them that we were looking to offload the most mundane kinds of transactions so that they could handle more complicated customer issues.

“We even talked about using AI as a coach for contact center agents – so letting them know we weren’t getting rid of anyone but that the platform might help them resolve an issue much more quickly,” he added. “We tried a few different ways to position it as ‘robots are not coming for your job’, but certainly that is the initial reaction... ‘I need to feed my family, so please stop what you’re doing’.”

Climbing the hype curve

The other panelists, who work for suppliers, noted that AI can, indeed, be a tough sell in some parts of the CSP business and stressed that AI is making its way through the infamous ‘hype cycle’, a term coined by Gartner. Indeed, the 2018 Hype Cycle for Emerging Technologies shows that AI hype hasn’t even peaked yet.

Augmented intelligence

When evangelizing AI internally, CSPs need to stress that the technology is not just a replacement for humans, the panelists noted. It can make humans’ jobs more interesting and worthwhile, and humans are still needed to understand data.

“At IBM we don’t like to say ‘artificial’ intelligence; we say ‘augmented’ intelligence,” said Janki Vora, Thought Leader in AI, IBM. “It’s not about taking away jobs; it’s making it better and faster. But at same time, IBM and some of our telco customers are investing in reskilling and retooling so that [displaced workers] can do something more complex… It’s a new era of different kinds of jobs that can be better and safer.”

Avai S, Director of the IoT, Big Data, AI/ML, Microservices Practice at Prodapt, agreed. “Employees should be made comfortable,” she said. “Without them we won’t be able to create value from the data.”

Russ Bartels, Director, Analytics, SDN & Network Automation,Windstream Communications, acknowledged that some attrition will result from implementing AI in customer service and operations, but like Avai he stressed that it’s crucial for CSPs to employ humans who understand the data.

“If someone’s great at what they’re currently doing and they’re willing to learn, odds are they are going to be able to continue working in another area,” he added.

Blueprint for success

Verizon’s Hamann said most CSPs don’t have a blueprint for success in implementing AI yet.

“We didn’t get far enough in our journey to prove out [the benefits],” he said. “We had some fits and starts and stops along the way, so we didn’t really get a chance to prove what we’re saying is true. In our case the skeptics kind of won…so it would be nice to have another crack at it because the technology has improved; the software has improved.”

Panel moderator Aaron Boasman-Patel, VP, AI & Customer Centricity, TM Forum, suggested that some of the onus is on suppliers to help operators develop a successful strategy for AI deployment. “What responsibility do you have to get it right?” he asked the panel. “Service providers will just get fatigue if they have bad experiences.”

Avai said it’s important for vendors to help CSPs establish not only short-term use cases for AI but also long-term business cases. “They are not the same thing,” she said.

Vora noted that success in other companies can help persuade CSPs as well.

“We may have very few CSP success stories that we are sharing, but there are companies out there that have AI in their DNA – for example, Alexa and Google Home. Those companies have embedded [AI] in their culture.”

Establishing trust

Given AI’s long road ahead, how can CSPs and suppliers establish the trust needed to make it a success story? Boasman-Patel asked.

“You’ve got to prove that it works,” Windstream’s Bartels said. “Your leaders need to believe in it and drive it, and you have to realize you’re not going to be successful every time.”

Verizon’s Hamann suggested that building AI trust is like elementary school science.

“You have a hypothesis and try to prove or disprove it,” he explained. “Whether you use AI to prevent network failures, help customers or create marketing offers, you have a control group where you’re not using AI and a group where you are using AI and see which one is better.”

He adds: “You need to try to be as dispassionate as possible. That way it’s a transparent process. You’re not cheerleading AI. You’re not trying to hype it or promote it. You’re just experimenting.”

Vora believes defining what trust means is key.

“A lot of organizations don’t have a definition that’s clear for everyone about what trust means,” she said. “Sometimes data is too guarded, but on the other end of the spectrum you need to have governance set up to make trust possible. The bottom line is that in order to trust AI systems, we first need to come to define what that trust definition is.”

Finally, Prodapt’s Avai pointed out that trust is two-sided.

“You have to have confidence in AI systems – that’s more of a technical trust,” she said. “The second is people trust – senior management and employees have to trust in AI.”