Artificial intelligence in the insurance industry

It is hard to avoid the subject of artificial intelligence when talking about the future of the insurance industry. But much of it is still at the idea stage or at least still being tested. InsureNXT talked to Dr. Andreas Becks, Head of Custumer Advisory Insurance at analytics software provider SAS, about the opportunities and risks and asked him for which application scenarios there are already market-ready solutions and what role sensor data will play in the future.

InsureNXT: In which areas and situations can insurance companies rely on artificial intelligence? How do insurance companies benefit from it?

AI is not always about high-end applications or robots that perform spectacular tricks. The core business of insurance companies is to assess risks – and this is already being done today with artificial intelligence. In principle, it’s about how machine learning can be used to create added value throughout the entire value chain, from product development to marketing and claims settlement.

InsureNXT: What savings potential does this offer for insurance companies?

Reducing the advantages of AI to cost savings would mean giving away potential. Instead, it is also about predicting risks and actively preventing losses or personalising offers based on a better understanding of risks and customers, dynamically adjusting prices and optimising customer care in the event of a loss. Automation and more precise, faster data evaluation create immense scope for improvement here. This creates new business models that meet the changing expectations of policyholders. After all, it is precisely the millennials who no longer want a rigid insurance policy, but rather want to insure certain stages of life. Increasingly, insurers therefore see themselves as partners for integrated solutions. A typical effect of changing consumer expectations is also the growing demand for “pay-as-you …” tariffs.

InsureNXT: Are AI-supported applications even ready for series production on a larger scale in Germany?

Beyond the hype term, machine learning already exists in Germany today in many “invisible” scenarios. Applications in the insurance industry start with product development, where the aim is to better understand risks and, on this basis, to create individual offers and dynamically adjust prices. In marketing, AI helps to create a better customer experience that integrates various communication channels. Claims processing, finally, benefits from automated sub-processes that allow insurers to quickly assess the probability of fraud, calculate the amount of loss and initiate the next steps. Cases can be quickly assessed and, for example, the driver of an accident-damaged car can be directed to the nearest auto repair shop, a tow truck or a taxi can be sent.

InsureNXT: An important limitation is the data situation – and Germany/ the EU are considered to be rather restrictive in terms of data protection. Is it possible to implement reasonable applications at all with the limited amount of data?

The data is not that limited. Insurance companies today collect completely new types of data that were not available before. This includes not only telematics data, but also, for example, vital data (on pulse, heart rate, movement, energy consumption), which a wearable collects. In addition, there are openly accessible data, and data can be purchased. Of course, the consent of the insured person must be obtained in accordance with the DSGVO. But in principle, consumers are quite prepared to provide their data if they receive advantages in return – such as a better tariff or a value-added service such as loss prevention.

InsureNXT: Will we be able to minimize risks in the coming years and insure more efficiently through better data? What role can Big Data play here?

Today we have a very large amount of sensor data. The good data situation in connection with IoT devices brings new insights that allow a more precise assessment of risks – and thus a fairer pricing. If a policyholder demonstrates a healthy lifestyle by means of voluntary transmission of vital data, the health insurance company can grant him/her appropriate bonuses. The evaluation of IoT data is already increasingly taking place in agriculture and can be used, among other things, to predict crop failures much more reliably – and then insure them.

InsureNXTSome start-ups pride themselves on being able to deal with damage within fractions of a second. How reliably does this work?

Start-ups such as Lemonade promise claims settlement within seconds – and in certain cases, they even honour this promise. However, these start-ups often do not have the best combined ratio, but will certainly learn from this over time or team up with experienced insurers. In any case, they are serious disrupters to certain core processes that shape consumer expectations. Because if I know that a provider will solve my problem immediately, I am not prepared to wait days or weeks with an established insurer. A better customer experience in claims processing is crucial.

As Head of Business Analytics DACH at SAS, Dr. Andreas Becks (SAS) is responsible for consulting on all aspects of analytical platforms. He leads a team of insurance experts, data governance experts and data scientists who advise insurance customers on how to use analysis to create value and drive transformation in a changing market.