Can AI Curb Insurance Fraud?

By Catherine Powell

Image courtesy of pix4free.org

While everyone complains about the high cost of insurance, what most people don't realize is that one of the things that causes rates to rise is insurance fraud.  Far from being a victimless crime, insurance fraud costs every policyholder.  That's because fraudsters bilk the insurance companies for hundreds of billions of dollars per year. The National Association of Insurance Commissioners (NAIC) recently reported staggering losses due to fraud last year that included $74.7 billion in life insurance fraud, $45 billion in property fraud, $36.3 billion in health insurance fraud, and $34 billion in workers comp fraud.  These losses in turn get passed along to each and every policyholder.  While the insurance industry spends a lot of time and money trying to combat fraud, the losses continue to mount each year.  Recent advances in artificial intelligence have brought some new tools to bear in the fight against insurance fraud that not only helps identify fraudulent claims, it can also curtail fraudulent activity.

While people power has been used in the past to spot fraud by looking for subtle patterns and telltale signs, new advances in machine learning have put this ability on steroids.  That's because AI can sift through mountains of information faster than an army of humans.  It also never calls in sick and it never needs sleep.  Insurers have recently crafted programs that can search for signs of: 

• people overstating the amount of damage done to a property.
• evidence of staged accidents or manufactured damage.
• falsified or exaggerated medical bills.
• fake claims using stolen information.
• duplicate claims.

Using predictive analysis allows insurance companies to assess the risk of fraud by assigning scores for each claim made and then comparing every new claim with millions of previous ones to determine how it measures up.  Any claim that is determined to skew from the norm can then be flagged for further investigation by a human being.  This not only saves time and money, it significantly improves accuracy when it comes to early detection of fraudulent claims.  

The insurance industry's use of natural language processing (NLP) can be a boon, since mountains of data can be crunched day and night, twenty four hours a day, seven days a week.  NLP has been the key to analyzing historical data of both legitimate and fraudulent claims which can then be compared to every claim filed in the past.  Such an undertaking would be impossible for human beings to accomplish. 

 

Image courtesy Pixabay

Advances in textual analytics and data mining bring new tools to the fight by automating the process of reading claimant responses, be they verbal or written.  AI is already helping to streamline both the adjuster process and claims processing by producing actionable insights that speed analysis, identifies inaccuracies, and provides better fraud detection.  Recent AI algorithms even allow searches of emails, text messages, and social media sites that are key in targeting the felonious activities inherent to identity thieves filing bogus claims.

How do you find the fraudulent needle in a claims haystack composed of thousands of claims per day?  In the past, this task was impossible to perform manually.  With the advent of AI, processing algorithms can automatically monitor the constant flow of claims to pluck out any suspicious ones in real time.  This not only solves the big problem of flagging potentially fraudulent claims, but it also saves insurers millions of dollars in losses and speeds payments of legitimate claims.

Cash for crash schemes involving everything from willing drivers and passengers to accomplices like towing companies, body shops and participating physicians have made staged car crashes a multi-billion dollar per year headache for insurers.  In the past, uncovering all the cogs in this kind of fraud machine was an extremely difficult and time-consuming headache for auto insurers and law enforcement from coast to coast. With the advent of AI,  it's now possible to spot fraudsters by sifting through news feeds, police and court databases, claims histories, and public records at the click of a mouse.  

Medical malfeasance used to be devilishly difficult to diagnose using traditional means.  By the time double dipping docs were caught it wasn't unusual for millions of dollars in overpayments or compensation for procedures that were never carried out to have been rendered.  Now with the help of AI, it's much easier to spot doctors who submit fraudulent claims by comparing their practice to databases of millions of their peers to quickly identify suspicious activity.

Workers trying to pull a fast one by blaming employers for injuries that weren't caused on the job can sometimes be thwarted by determining whether these employees reported the same occurrences multiple times.  The same analysis can be used to save retailers from having to pay dubious slip and fall claims by determining whether the claimant has reported such an event on more than one occasion.

Crime never sleeps.  In order to stay one step ahead of fraudsters, the insurance industry needs to diligently pursue every avenue open to it to keep fraud at bay.  One of the most promising avenues is AI.  Unfortunately, criminals who commit fraud have access to artificial intelligence too, which could lead to a kind of AI arms race.  As the anti-fraud fight rages on, insurers and tech companies are working to develop more robust AI tools designed to help insurers, investigators, regulators and  law enforcement agencies combat the growing scourge that continues to hit every American in their wallet.  This is one cold war that we can't afford to lose.

Catherine Powell is the owner of A Plus All Florida, Insurance in Orange Park, Florida.  To find out more ways to save on flood insurance, check out her website at http://aplusallfloridainsuranceinc.com/