The insurance industry has evolved more in the last five years than it has over the previous 50. Advances in technology upended the standard top-down legacy model such that property or vehicle owners can now download one of dozens of apps and sign up for a policy in minutes. This ability to take full control over their insurance policies while cutting out the traditional gatekeepers has changed the way the game is played. Yet, the advent of insurtech and the deep personalization it brought are still not true disruption. Artificial intelligence (AI) is the actual disruptive force in insurance. The insurance industry's philosophical underpinnings may perhaps be oversimplified as follows: Party X insures Item/Asset Y by giving money to Insurer Z with the expectation of some level of compensation promised should something adverse happen to the insured item. That's all well and good, and it was this simple formula on which the concept of insurance was based for centuries. But today, with the advent of AI technologies, insurance providers and policyholders alike have access to oceans of data and machine learning-based predictions and insights. This visibility was previously unheard of and has been revolutionizing standards of practice across many legacy industries, with insurance at the forefront of the revolution. Many in the insurtech sector believe AI will radically upend most, if not all, lines of business in the property and casualty insurance space. Why? Because most P&C insurance is rooted in a promise to compensate humans for their errors. AI represents the technology antidote to such errors by anticipating them and, increasingly, correcting them before they even occur.

Legacy weaknesses

Since its inception, legacy insurance has been plagued by fissures that were seen as necessary evils. The U.S. auto insurance sector, for example, sees more than 38,000 traffic-accident deaths a year, according to the National Highway Traffic Safety Administration, with a price tag of $55 billion in medical and work-loss costs for those fatalities, says the Centers for Disease Control and Prevention. What's more, 346,800 homes are destroyed by fire each year, says the National Fire Protection Association, with a cost of $7.3 billion in direct property damage. The National Council on Compensation Insurance reports that each workers' comp claim costs more than $81,000 a year. The FBI reports that insurance fraud zaps more than $40 billion a year. These losses increase annually with no sign of slowing. But with the visibility, specificity, fraud reduction and customization offered by the new data-driven AI reality, many of these insurance problems can be tempered. This can save customers money while allowing insurers to continue providing coverage without a constant existential threat to their bottom line.

Core insurance practices

Auto. About 1.3 million people worldwide die every year as a result of automotive crashes, according to the World Health Organization. Replacing human error with precise, predictive decision-making constitutes the core of AI's contribution to the consumer experience. To that end, it will revolutionize driving and save lives. Specifically, self-driving automotive technology is becoming more sophisticated, despite the fact that some heady predictions of fully autonomous, consumer-ready vehicles were a bit premature. It is, however, only a matter of time until they become the norm. When that happens, the human error responsible for the vast majority of auto insurance claims will become virtually obsolete, replaced by AI-powered "smart cars." The new era of personal transportation and mass transit will by definition carry humanity into an age in which accidents and breakdowns occur at a fraction of the rate they do now. A report by McKinsey and Co. estimates that widespread adoption of self-driving cars will eradicate a whopping 90% of automotive accidents and almost $200 billion in damages and medical costs in the U.S. annually. Auto insurance claims will plummet in parallel. It must be said that replacing drivers with AI will not completely solve the accident problem. Driver error is indeed the "final failure" in more than 90% of crashes, according to the Insurance Institute for Highway Safety (IIHS). To avoid such collisions, automated vehicles will need to be "specifically programmed to prioritize safety over speed and convenience," the IIHS says. Home and property. AI has the potential to profoundly curtail home and property damage and likewise help reduce insurance claim costs. Consider a recent LexisNexis water loss study that found the Internet of Things (IoT) and smart home technology could reduce water insurance losses by 90% by 2025 in fully connected homes with water and fire prevention devices. Today's AI tools can detect water leaks well before they happen and shut off the water supply automatically. And far beyond robotic vacuum cleaners, the future will see life-sized humanoid and other robots with the ability to pick up a fire extinguisher, walk over to the Christmas tree, and blast the sparking, defective colored bulb that might other-wise have ignited and brought down the tree (along with the entire house). Workers' compensation. In his 2020 book "The Future of Insurance: From Disruption to Evolution," author Bryan Falchuk illuminates flaws in the current workers' compensation system: "The biggest problem I see in Workers' Compensation is the asymmetry of information. Carriers do not really know what the exposure is since they cannot see the true risk employees face day to day. They send engineers on site to observe things, and employees all follow the safety rules when representatives from carrier or safety engineers are on site for observation." "However, you never know what these same employees do when the carrier's representatives aren't there. Safety equipment, such as hard hats, may not be worn, areas with restricted access may be freely walked through as short-cuts, and warning signs, lights and alert tones may be ignored. Or carriers may presume as much and bake into their rating an assumed non-compliance rate for these safety measures that are being followed. This means carriers are rating the policy somewhat blindly, potentially giving discounts that should not be given, or perhaps overcharging for risk that is not there." But now, companies such as Safesite, Pillar Technologies and Guardhat are correcting these blind spots and taking the guesswork out of workplace safety and compliance by using AI-powered cameras, sensors and protective equipment to foster safer workplaces. Safesite and its parent company, Foresight Group, report that their technology can reduce the frequency of policyholders' workplace safety incidents by an average of 31%. Insurance Fraud. Fraud in the insurance world is the best example of Donald Rumsfeld's famed "known unknowns." Insurers know it exists and can sometimes detect it, but the true amount of fraud in any given line of business in any given year is not known. While there are many excellent insurance-industry groups like the National Insurance Crime Bureau available to help combat fraud, there is no solution or suite of solutions that has been shown to eliminate it. AI promises to be a major improvement in both fraud detection and fraud prevention through its pièce de résistance: pattern and anomaly recognition. As one example, FRISS and Clearspeed are combining fraud detection technologies to help companies like Seguros El Roble optimize their claims process and significantly reduce fraud. This empowers adjusters to make more productive and accurate referrals to insurer special investigative units (SIUs), the supplemental departments that work with claims and underwriting to investigate potential fraud. Third-party administrator (TPA) services provider CodeBlue replaces the conventional First Notice Of Loss process with what they call a "First Notice and Response (FNAR)" service. This innovative, AI-based claims strategy features virtual collaboration between the policyholder and CodeBlue's claims specialist. The company's response process eliminates unnecessary demolition and reconstruction, limits unnecessary travel time and creates efficiency, allowing adjusters to settle claims more quickly. It also reveals fraudulent behavior by capturing an image of the claims site before unscrupulous contractors can get in and begin unnecessary repairs.

Insurance reconstructed

Insurance at its core attempts to understand and price peoples' conduct (human behavior or the human element in industry jargon) so that it can protect against careless or malignant behavior. Cover Whale, for example, is an insurtech company that uses data analysis and AI to insure truck drivers. The idea is that a better view into typical and atypical trucker behavior can allow more precise insurance coverage plans, and in tandem, lower premiums. Koffie is another example of a trucking insurance business that tracks and analyzes real-time truck fleet data. In doing so, it incentivizes safe behavior. Like Safesite and Pillar, these commercial automotive tech insurance companies are taking this data and turning it into real-time micro-services that coach drivers and help them avoid making careless mistakes like falling asleep at the wheel, backing up in a tight space with unseen obstructions, or spotting a car approaching at an unsafe speed and alerting them on the best way to avoid an accident.

First-person perspective

Earlier this year, I had the unfortunate experience of watching as a wild driver zoomed past me on a two-lane canyon highway near my home. My kids happened to be in the car with me when this happened. It was a foggy day, and the traffic had slowed due to the poor visibility. About two miles down the road, I witnessed that same car perched in a ditch on the side of the road, its driver dazed and lucky to be alive. As I explained this progression of very-bad human decision-making to my horrified children, I made sure they understood the risks of reckless driving. But I also shared that eventually, in their lifetimes, they would not need to worry about this risk because their future cars will have built-in computers programmed to help them avoid such dangerous behaviors and collisions. Artificial intelligence promises to improve our living standards through augmenting behavior and reducing error. This will in turn save insurance carriers money and people money, pain, time and emotional distress. AI can fulfill the promise of good technology to make life better and easier. In the past, insurance helped improve business and social practice by acting as a regulator of behavior predominantly through price (premiums) with some, but not much, behavioral impact. Today, artificial intelligence is ushering in the expanded role of insurers. AI adoption in insurance will increasingly help consumers and businesses prevent and resolve problems on their own. The prime impact will be lowered premiums in existing lines, freeing up capital to allow insurers to do what human insurance talent does best: discover and provide risk finance solutions for new problems amidst humanity's ever expanding pursuit of new knowledge. Daniel Turgel ([email protected]) is CEO of the insurtech companies MySmartInsure and MindHome. See also:

• AI: Insurance, business and liability considerations
• How insurers can build confidence in their AI
• Consumers voice unease with AI-only home & auto claims