Cutting Human Error: AI in Insurance

Right now, claims processing is largely a manual function. This makes it prone to human error. Error means less efficiency and less efficiency means higher costs for the insurer. Big data allows AI to be effective and more accurate. As more data becomes available, AI will become even more useful in uncovering suspected fraud that previously was undetected.

V7 Labs reports that because the claims management process is largely paper-based and rarely start-to-finish digitized, it can eat up to 50%-80% of the premiums’ revenues.

So, what is driving the adoption of AI? Connectivity is one reason. From the cars we drive that have onboard telematics, tech watches, fitness trackers, to smart home devices, insurers are in a position to automatically be collecting comprehensive data on their customers. From here it is just a matter of blending this information into the claims management process to make them less prone to error and quite a bit faster.

Drones and Insurance

Machine learning algorithms can swiftly scan all the incoming data, interpret it instead of insurance agents, and provide faster settlement results. This process will never replace humans as there needs to be an agent on the other side of this to approve the outcome.

Positives

AI can decrease the labor-heavy and dangerous inspection tasks allowing for a more effective claims management process.

You may not believe it, but property insurance adjusters get injured 4 times more often than construction workers! Who would think operating a crane would be safer?

Keep our adjusters Safe!

Drones are becoming an option for property adjusters to examine roof damage and provide an estimate. These cameras with wings allow adjusters to inspect a natural disaster area or industrial equipment such as oil pipes.

Save billions annually

According to the FAA, by 2022 it is expected that almost 3 million drones will be buzzing around, with 450,000 of them specifically tied to commercial applications. The insurance industry has been one of the earliest adopters of this technology. The functions for them have applicability across the entire value chain. It has the potential to help save billions of dollars annually by improving risk management, resource efficiency, fraud reduction and employee safety.

Challenges

The FAA requires that the operator of a drone and/or the visual observer be close enough to maintain a constant line of sight with the drone. This could limit the use of remote inspections for an investigation.

What happens if the drone causes damage?

In addition, there is risk of liability in the event of injury or property damage because of a drone. The courts may recognize a ‘tort duty’ for the operator to be using ‘due care’ to prevent such damage or injury to themselves or others. Regulations are being developed as we go, and new federal or state rules may continue to shape the landscape of what this “duty” portion of claim negligence can mean.

Privacy Concerns

Courts have been hesitant to uphold trespassing claims that involve drones or other aircraft that are operating below the airspace threshold which can interfere with the use of the land by a property owner. Further, a drone operator can be liable for retrieving a drone that failed on someone’s property. Physically entering private land can still be considered trespassing even if the drone use was legal. In the case of a risk assessment flight, it seems likely that lawmakers will expect prior permission from the owner of the property.

The future of drones is looking bright. As a developing technology already taking up 17% of the total usage by the insurance industry, we expect this to make big strides sooner than later. Tied to a carriers’ data system, fraud investigations with real-time data linking together from multiple sources will become the standard.

AI-based systems can process:

• Geospatial data collected by satellites

• HD video or imagery, shot by a drone

• Data sets, including temperature, pressure, object position, and more

Use of AI in Insurance

We are all too familiar with the statistics of the annual cost of fraud. It is a pestilence in our society and yet the fight to stop it may never end. The rapidly outdated systems of detecting elaborate fraud schemes must evolve if we are to keep up.

Case Study #1

Let’s look at an example from Tokio Marine, an auto-insurer based out of Japan. They recently implemented a computer vision system for examining and estimating the damage on vehicle claims all AI based. The AI is trained on databases with millions of photos of car damage as well as human appraiser decisions, and the algorithms learn from experience by analyzing a large variety of different examples.The turnaround time on an automotive damage claim in Japan is between 2-3 weeks. With this technology they anticipate they will be able to do in minutes what is currently taking days.

The company will use the AI to understand the full range of available repair decisions, including recommended repair, paint, and blend operations, as well as the labor hours required. The AI evaluates the damage to a vehicle, based on photos provided by repairers, appraiser or policyholders. Any red flags appearing during this real-time appraisal can be noted for additional investigation by the adjuster.

This means quicker claims processing from start to finish with less room for error. The AI they are using has already processed over a billion dollars’ worth of automotive claims in the UK, Poland, and France.

Case Study #2:

Machine learning can identify patterns that are out of the ordinary or recurring and make the AI and its related systems a powerful tool. There is a Turkish insurer called Anadolu Sigorta. This company had historically reviewed each application for fraud potential manually. This process took up to two weeks for each application. They managed up to 30,000 applications per month.

Obviously, this laborious and lengthy process was full of missed opportunities. They decided to turn over some of the screening duties to a new AI platform that would allow them to use a predictive system to identify fraud in real-time. In the first year they increased their ROI 210% and attributed a savings of almost $6 million to the fraud detection system.

Ultimately what artificial intelligence is trying to do is make the analytical process for adjusters and investigators easier. The objective is to generate good referrals. Flags that are more accurate, speedy, and applicable. This quickly available information means the investigation of potentially suspect claims can be addressed in a faster more resolute manner.

The Goal of AI

AI lacks some fundamental tools it can never obtain; thus, it will never overtake human interaction. But it can assist humans to be faster, more knowledge and increase productivity. Imagine the ease of creating a research paper with the use of google rather than scouring through a textbook. AI sifting through data for you is the equivalent of technology that is being used in nearly all industries.

Even if your company doesn’t directly rely on AI to advance their claims management, we are one of the most qualified tech-enabled companies in the risk mitigation arena. Outsourcing of your tedious tasks can give you faster turnaround times on your claims and more results.

Ethos strives to put an emphasis on technology for the ease of our client. Use AI and drones where it makes sense. Reach out to us if you need a hand in the process.