How can AI be safely integrated into an insurance workplace?
How can AI be safely integrated into an insurance workplace? | Insurance Business America
Technology
How can AI be safely integrated into an insurance workplace?
This technology has the potential to grow productivity within a company
Technology
By
David Saric
As more insurance businesses adopt or contemplate the benefits of using AI technologies in the workplace, it is important to consider custom platforms that are created specifically for a business, an Applied Systems leader has suggested.
“It’s definitely worthwhile to spend some time looking at services that expose models that are unique to you, and that have favorable terms and conditions to make sure that you retain copyright,” said Tanner Randolph, chief information security officer at Applied Systems.
However, businesses should be mindful of the limited scope of these AI platforms and how they should be used in tandem with human intelligence, according to Randolph.
“AI doesn’t actually know anything; it predicts what it should use as a next word,” Randolph said. “It’s really just a prediction model with a lot of data power.”
In an interview with Insurance Business, Randolph spoke about how AI and automation, when used correctly, can be used to bolster productivity. He also debunked the myths around generative technology and why the insurance sector is slow to adopting new technologies.
“We’re seeing massive productivity gains”
One of the main advantages of AI that has been observed across multiple industries is the ability to replace manual and time-consuming duties that can be handled by automation.
“Some of our internal testing has shown that there is a 40% to 60% increase in productivity,” Randolph said.
This is in relation to generating new content and ideas, which can be particularly useful for emails or refining the content within them.
Applied Systems tested this out on marketing material and found that, with certain guardrails in place, AI can create highly creative outreach programs for customers in as little as a five-minute interaction with a generative solution.
Although, in order for AI to be leveraged properly, businesses should not solely rely on this technology to produce content or results independent of human intervention.
“Internally, we’re seeing a lot of advantages of starting with [AI], and not necessarily finishing with it, but making sure that you’ve got a human in the loop to get the most out of it,” Randolph said.
Furthermore, other day-to-day duties, or as Randolph remarked, “the ticky-tacky stuff” can be effectively eliminated from an insurer or broker’s list of tasks, leaving room to place more emphasis on relationship building.
“These systems have the potential to keep up with your workflow, and that’s a game changer for a lot of the industry because it is so personal, at least for independent agencies.”
Debunking certain widespread myths
With the introduction of any new technology, certain preconceived notions or falsehoods are bound to circulate and cloud people’s judgements.
For Randolph, the most pressing myth related to AI that needs to be debunked is the notion that AI is “intelligent.”
He pointed to the Waymo self-driving vehicles that have been making the rounds in San Francisco, stating that the easiest way to stop one is to put a cone on top of it.
“In security terms, you’ve just impacted its availability, and it doesn’t know how to operate like that,” he said. “Once you deviate from the norms of the AI model, it has no idea what to do.”
AI is only supposed to work within the parameters that is was programmed to be successful in, thus, it can not be relied upon to provide an intelligent response or solution to a problem outside of that specific capacity.
This is evident in the security functions of these programs, which are only as effective as they were intentionally designed to be.
“AI wasn’t designed with security in mind,” Randolph said. “You’ve got to have controls to make sure that data can’t egress out of that model, which is possible, both within traditional data and machine learning and in generative AI.”
Why the insurance industry is slow to adopt new technologies
Some people view the insurance industry as moving at a glacial pace, but according to Randolph, this is more widespread.
“I’ve worked in multiple fortune 50s in both retail and health insurance, and inside of the defense industry — they all move at a pretty slow pace,” he revealed.
This is largely driven by a more cautious approach to investing in new technologies that might not make it through a consolidation phase of three-to-five years.
“Most of these companies have 50- or 100-year time horizons, so you would expect them to be relatively risk averse,” Randolph said.
“Personally, I wouldn’t be overly thrilled if I knew that my insurance company was being extremely risky with the funds that I was depositing there.”
Although, when speaking with venture capital professionals and advisors, Randolph found that many of the big players within financial services, including insurance, are moving quite quickly to adopt certain AI capabilities.
In this article for @InsuranceBizUS, our EVP and Head of Insurance Vikas Bhalla predicts that #GenerativeAI solutions will take off across the insurance landscape in 12-18 months. Discover the potential use cases: https://t.co/Cf04MKaYen
— EXL (@exl_service) August 4, 2023
“I think everyone sees it as a new version of cloud computing,” he said. “I was around when Amazon Web Services became a thing and witnessed the uptake there — I think we’re seeing a similar pattern here.”
What is your view on uses of AI in insurance and Applied Systems’ Randolph’s insight? Leave a comment below.
Related Stories
Keep up with the latest news and events
Join our mailing list, it’s free!
