How to use Gen AI to reduce your compliance burden
Insurers face increased pressure to keep up with a myriad of regulations, guidelines, and evolving industry standards. How to cope with this frenzied pace of regulatory change?
Enter generative artificial intelligence (Gen AI).
Gen AI can help insurers analyze and interpret regulatory requirements with much more agility than before, says David Mamane, financial services senior analyst at RSM Canada. “[Gen AI] is enabling [insurers] to consume large amounts of text data within guidance and regulation,” he reports.
Insurers can now input volumes of documents into Gen AI, which then helps insurers summarize, research, and interpret the regulatory information. Insurers can now query the Gen AI on the contents of the information.
The new tech can reduce the overall time spent on pouring over those lengthy documents. Also, it can generate responses to questions about how certain regulations might affect the business.
Mamane’s seen newly incorporated P&C carriers, as well as companies with new business models, use Gen AI effectively to comply with Canadian regulations, For example, Gen AI can summarize lengthy and detailed documents such as the Insurance Act, or other provincial or rate filing guidelines. This can be exceptionally helpful for new markets, he says.
“When you think about the Insurance Act and various regulatory guidance that new carriers have to comply with, it can be a lot to digest and a lot to understand,” Mamane says. “Especially for new carriers investigating and launching innovative business models that don’t necessarily have a precedent in the Canadian market.”
When new carriers want to navigate long-standing regulations with Gen AI, the key is to use it solely to consume and understand those practices or guidance. Then, the AI can be trained to refer to those documents in its future outputs.
For existing carriers, Gen AI can be particularly helpful to keep up with new and rapidly changing regulations.
Another possible use case for Gen AI is to produce answers or solutions to regulatory requirements.
“This is an evolving landscape, constantly,” says Mamane, “[By] allowing carriers to navigate those rate requirements — allowed rating variables; disallowed rate variables; new ways to price risks — you can get some really good assessments out of Gen AI tools.”
For example, he says, “a lot of our clients in the insurance space have asked [Gen AI], ‘What does the Alberta auto reform mean for me and my business?’”
Speaking of Alberta auto reform, insurers could use Gen AI to research examples of how comparable changes to insurance models have played out in other jurisdictions. Or, as Mamane puts it: “’What other countries, what other regimes, what other states have gone through regulatory reform that looks like this?’”
Canadian auto insurers are still waiting to hear from Alberta about its much-anticipated auto reforms. In the meantime, posing a query about Alberta’s options to Gen AI might give insurers additional context on how auto reform options have fared elsewhere.
Human touch
Asking Gen AI to summarize regulatory material might be considered relatively low-stakes. But asking it to answer specific queries about regulation must be done with more care.
Firms must use a fine-tooth comb before implementing outputs from Gen AI, says Mamane.
The key is to ask specific questions and provide as much context as possible (while being mindful not to share any client data or proprietary information).
“Setting the context, setting the stage, can really avoid or minimize hallucinations within Gen AI tools,” Mamane says. “Then once you’ve set the context, that’s not usually enough; you have to consistently and iteratively validate.”
By iteratively prompting the Gen AI with new queries, insurers can systematically refine and adjust the outputs and improve its responses.
The iterative process doesn’t take long, Mamane says.
“We’re talking minutes or hours to do….What normally would have taken a human hours, weeks or days to source all this material, read through it, render an opinion, you’re 1722386206 spending a few extra hours to do that vetting process. It’s still saving you a tremendous amount of time.”
Insurers can also vet the output they receive by asking Gen AI to directly quote the documents or specify where the information comes from.
Talent in crisis
Workforce challenges plague the industry as a whole.
Mamane’s noticed the “experts that have historically known this regulation like the back of their hand are becoming scarcer and scarcer.”
Whether due to rapid retirements or budget constraints, talent shortages mean teams are less likely to have built-in regulatory experts.
Although Gen AI by no means replaces human-decision making, it can become a tool for efficiency.
“Make sure you have those human peer-reviewers that go through [the outputs],” says Mamane. “Still, an AI-enabled human in this case will be a much more powerful problem-solver.”
Feature image by iStock.com/demaerre
