Leading insurance executives provide their insights on AI project roadblocks, how to overcome them, and overall best practices.
In our recent webinar with EXL, Conversational AI Can Deliver a Superior Customer Experience (if you missed the original session, be sure to watch the replay), we asked our audience to participate in several polls regarding attitudes toward AI technologies and deployments. The results provide an interesting point-in-time snapshot of how enterprise buyers and decision makers are considering AI as part of their future strategies.
For the first poll question, we asked participants: “Is Conversational AI in insurance just ‘hype’?” Among those that responded, 93% of participants said AI is not hype, and that the technology helps insurers deliver better customer experiences. As for the remaining 7% that labelled AI as hype, research shows they are not alone; according to a recent KPMG study, 74% of executives believe the use of AI for business is currently more hype than reality.
Our panel of experts shed light on why AI may sometimes feel like an overhyped technology in insurance. Ankor Rai, Chief Digital Officer at EXL, agreed that sometimes “AI works beautifully in the lab,” but “when you take it into the real world it doesn’t.”
Instead of jumping to conclusions about the technology, Rai encouraged organizations to instead reflect on their own expectations of AI. Rai highlighted how we often expect AI to solve the hardest problems first. However, like how children begin school in first grade and work their way up, AI technology needs to be trained and developed, as AI’s improvement is “exponential.”
Jamie Stevoski, Customer Experience and Digital Enablement Manager at insurance provider IAG, also posited that feelings of hype stem from high expectations for AI, especially when paired with insufficient project planning. “Organizations get quite romanced by the robots,” said Stevoski.
Implement the Right AI Project Plan
Stevoski also said that organizations often “try to get to the end-game overnight.” When companies rush into their AI deployment, they often do not reap the ROI on the technology. Instead, Stevoski recommended that organizations create an AI project plan with measurable goals. “It’s quite important to break that plan down, break down what your horizons [are],” he said.
Perceptions of AI hype may have little to do with the technology itself, but rather the data with which it interfaces. “The area we have to start thinking about is data, because you’re not recreating data for each AI system. You actually have one data set that the AI systems are looking to interface with,” said Tejash Patel, Vice President and Chief Architect of insurance provider Guardian Life.
Patel said that “the ability for AI systems to communicate is going to be a really key area and I think that’s where the hype will turn into reality, but we’re still early stages.”
AI Deployment Roadblocks
For the second polling question during the webinar, we asked our audience to vote on the main challenge that hinders AI initiatives. The responses to this question were a bit more varied, with 46% pointing to a lack of usable and useful data, while 31% said legacy insurance systems and the associated integration effort were the main obstacles.
Of the remaining participants, 14% said the main challenge was that available AI platforms were not mature for enterprise adoption, and 9% said unsuccessful prior chatbot and point-automation initiatives hindered future AI success within their organizations.
Fortunately, the webinar panelists provided insights and best practices for how to overcome AI deployment challenges, whatever they may be.
Overcome Organizational Obstacles
AI requires data in order to work effectively. However, not all organizational data is necessarily sufficient to enable AI to function seamlessly. As Patel said, “data has to be of good quality.” For example, he mentioned how companies like Guardian can make improvements to their data by simplifying the “definition of product classes.”
Stevoski also argued that, in addition to using good data, internal processes need to be aligned with companies’ AI strategies for projects to be successful. To do so, Stevoski suggested that organizations should build a customer journey map to act as a blueprint. "What’s really key is ensuring that your customer experience has a clear view end-to-end,” said Stevoski.
For example, at Guardian, AI-powered customer support is available company-wide. Hugo, Guardian’s Amelia-powered virtual agent, “is going to be the digital assistant across the enterprise,” said Patel. “The AI construct needs to be at the foundation versus at one end of the pillar.”
If organizations implement AI without addressing these organizational obstacles and refining their business processes, the result will be a more frustrating customer experience. As Stevoski said, “In order to apply an AI-like outcome, you need to refine, you need to innovate, you need to really challenge yourself around what you currently do in the current climate and really evolve.”
Nurture Customers’ AI Adoption
“Just because you build it, doesn’t mean that they will come,” Stevoski said of AI projects. Regardless of how innovative the technology, companies will not see ROI from AI unless their customers actually use it. Rai highlighted how, for customers to work with AI, companies need to be mindful of when the technological fails. “You want to proactively figure out when you are hitting an issue and route it to a human [when necessary],” he explained.
Guardian’s approach to increasing clients’ AI adoption is to not force the technology upon the customer. “It’s more of an option for them,” Patel explained. He said Guardian’s method is about “introducing [AI] through [their] website as an opportunity.” By presenting AI as an easy-to-use, efficient alternative to picking up the phone or sending an email, customers will likely become more comfortable with using the technology.
To hear more AI and insure-tech insights from the webinar panelists, be sure to watch the full webinar replay.