surreal image of arm coming out of yellow wall holding the receiver of an old phone
PHOTO: Elena Koycheva

Artificial intelligence (AI) is providing some payoffs in customer experience (CX), but provides the most value when part of a comprehensive CX strategy, according to a newly released MIT Technology Review report sponsored by Genesys, "Human and Bots: Tension and Opportunity."

The report found CX leaders and other large companies making major investments in AI to improve their efforts. “Customers worldwide expect a high level of service that can only be provided with the support of technology.”

The technology itself is used primarily to enhance efficiency, though leading CX firms use AI “to bring a deeper level of customer understanding, driving customization and a personalized journey.” Nine-in-10 survey respondents said since incorporating AI, they had seen measurable improvement in the speed of complaint resolution. Eight-in-10 said AI had enhanced the number of calls they were able to process.

AI can provide automated responses to an increasing number of and increasingly complex frequently asked questions, thanks to machine learning and human fine-tuning of responses. As a result, customer revenue and customer satisfaction are improving, with 80 percent of survey respondents seeing measurable customer satisfaction improvement, according to the report.

Though AI adds to operational costs, increasing customer sentiment analysis is a critical factor for leaders in CX technology deployment, the report stated. “By investing in tools that drive deeper customer understanding, they are able to make smarter decisions.” Nearly all of the survey respondents who had applied AI-enabled technology at some touchpoint in the customer journey report seeing immediate impact on CX, the report added.

Related Article: 10 Ways AI Helps Improve Customer Experiences

Where AI Is Showing an Impact on CX

While it's still early days yet with AI, success stories have started to emerge. 

  • Rakuten, a Japanese ecommerce firm, has been able to move three quarters of inbound calls to chatbots, greatly increasing the volume of calls the company can handle.
  • Canadian mobile telecom carrier Telus was able to offload 40 percent of calls to chatbots, improving service levels and cutting customer wait times from an hour to 2.5 minutes, improving operational efficiency as well as CX. Customer service representative satisfaction also improved.
  • Alibaba, the Chinese version of Amazon, used AI-powered chatbots on the company’s Tabao site to handle more than 93 percent of customer queries in 2017, which company officials say would have required 83,000 human agents working around the clock. Alibaba now offers these chatbots as a standalone service.

Other CX payoffs the report noted included automatically moving a caller to the right human customer service representatives, but it didn't provide any details on success rates.

Related Article: The Future of Customer Experience Is AI: Are You Ready?

No Silver Bullet

Despite the success that some have had with AI, it is not a solution in and of itself, cautioned Omer Minkara, vice president and principal analyst for Aberdeen.

“There’s a misperception of what AI can and cannot do,” Minkara said. “One of the ways that AI is presented is as pure automation, with the idea that you can get away without humans.” While AI is fine for simple tasks, like finding an account balance, or even somewhat more complex tasks, like providing types of accounts available, it has yet to advance to the level of being able to handle more complex tasks, like handling an insurance claim, Minkara said. Though AI might be able to record some basic information, like time of claim, name of claimant, etc., insurance claims as well as many other complex interactions still need human interaction. 

According to a 2018 Capgemini survey, 55 percent of consumers prefer to have interactions enabled by a mix of AI and humans. The most important area to have the human touch is for “high-consideration” transactions — typically more expensive transactions, like buying a car. Forty-five percent of respondents to the survey preferred human-only interactions for high-consideration interactions, as opposed to 30 percent for low-consideration ones.

“Sometimes it is about the situation in question, rather than just the cost,” said one respondent. “I could buy a car online via a virtual assistant, if someone recommended it, even though it is an expensive product. But if I have a car accident and need to deal with my insurance company, I would find a human interaction more assuring.”

Yet, despite the effectiveness of human interactions in many of these situations, AI providers often present the technology as an opportunity to reduce or eliminate the workforce, Minkara said. “That’s not the case. When companies understand this, AI is really good for augmenting human intelligence.”