It’s crunch time for insurance companies. Investment yields saw a sharp decline while combined ratios have increased steadily over the past three years. This has insurers asking themselves, “What can we change operationally to improve our economics?”
Fortunately, concrete answers are now available thanks to artificial intelligence — especially the variety of AI technology focused on knowledge and textual information. There are three areas in which this technology is delivering unprecedented productivity gains and insight and leading to deep changes in how insurers do business.
Why Artificial Intelligence is a Great Match for Insurance
Insurance is a very manually operated and paperwork-intensive industry. Both underwriting and claims management processes revolve around the retrieval or production of information from documents —such as a policy or a claim package — that capture the specifics of the insured’s case. These are time-consuming tasks that are highly dependent on human operators and, on average, represent 14% (health) to 20% (P&C) of the combined ratio.
AI (especially NLP/NLU) technologies have proven particularly effective in reducing the time and effort necessary to complete these tasks by reading your important, text-heavy documents with human-like accuracy. This enables you to locate and extract key information necessary to complete your tasks. We call this intelligent automation.
This has proven to significantly reduce time spent on underwriting and claims management from hours to seconds. As a result, insurance professionals can refocus their time on high-value areas of their jobs (i.e., making final determinations), rather than menial time-consuming tasks.
The following breaks down three of the more common ways in which AI has shown to impact the insurance industry.
#1 Area: Reducing Leakage
The number-one expense item for insurers is loss. With premium leakage routinely estimated in the billions (see: Accenture’s Claims at a Crossroads or Verisk’s The Challenge of Auto Insurance Premium Leakage), insurers must actively locate and plug these leaks. AI is a great resource for doing so.
For example, AI has proven valuable to the claims workflow by detecting possible fraud patterns, such as inconsistencies in accident descriptions. This helps to highlight key details in cases that are more difficult to for claims handlers to identify and prioritizes which cases to investigate.
AI has also been effective during the policy review process. Not only can it identify and detect over-exposures, misalignments and other red flags, but it can compare policies to others to better pinpoint potential areas of leakage.
#2 Area: Augmenting Underwriting
Risk evaluation is essential to the job of every underwriter. It is the process by which underwriters grade prospective customers on select risk factors and mitigators to determine the level of coverage they are willing to provide. Traditionally, this is done manually via analysis of third-party risk reports.
However, with AI, you can quickly and consistently extract each of these indicators, then report an overall, evidence-based, risk grade to the insured. This enables risk engineers to allocate more of their time directly toward the verification of highly com¬plex cases. In return, underwriters can turn around new policy quotes faster — a key competitive advantage.
When applied to the review of existing policies, AI can analyze clauses and compare them to reference policies to flag misalignments and excessive expo¬sure. This enables underwriters to more easily flag and prioritize clauses they may need to renegotiate in the future.
#3 Area: Accelerating Claims
Claims workflows are dependent on proper information flows, but claims handlers continue to spend a disproportionate share of their time on ancillary informa¬tion grooming tasks. AI can alleviate much of this issue. For instance, when dealing with automotive claims, it can:
- automatically review claim packages to evaluate their complexity and route them accordingly,
- recover essential aspects of accident descriptions to support liability determination, and
- give suggestions based on the facts of the case (e.g., which injuries were incurred, what medical tests or treatments were applied, etc.), distinguishing between current prognosis and past medical history.
This all leads to an accelerated claims workflow, which frees up time for claims handlers to focus on the most important of their work: decision making. And it’s not just output where companies are seeing a difference, but customer satisfaction as well. Claims settlement speed is regularly cited as a leading contributor to customer satisfaction.
That’s it for today, but we stay tuned for our blog next week where we show you how to build a business case for applying cognitive to your own business.
