Approximately 2 minutes reading time
Algorithms are more frequently used by insurers, for example for policy acceptance and claims handling. Nonetheless, the insurance sector is not a frontrunner when it comes to the development and use of algorithms.
The current applications of algorithms are often based on fairly simple (business) rules that are easy to interpret. During acceptance, for example, a number of checks are carried out that are used to calculate a final score. When an insurance claim is filed, a risk analysis is carried out based on data from different sources. Although this analysis is more sophisticated than it used to be, it is still very easy to understand.
The advantages of algorithms
Slowly but surely, machine-learning is being used to develop algorithms that are less transparent and not so easy to explain. A number of insurers have already started experimenting with algorithms that make a fully automated real-time damage assessment on the basis of photographs. This eliminates all human involvement in the claims handling process.
We can also see the advantages of algorithms in another area: premiums and products can be personalised based on lifestyle, behaviour and/or use of certain products. A simple example would be a young mother driving between two small townships paying a significantly lower premium than a person in his/her twenties driving through the centre of Amsterdam. Instead of being based on where somebody lives, the premium is based on where they normally drive.
Pay-per-use offerings will also become more sophisticated in the coming years, and it will be possible to calculate specific risks much more accurately, especially as the IoT becomes more widespread. This not only has implications for such issues as solidarity – will some people become non-insurable? – it also raises the issue of whether or not algorithms discriminate unfairly.
Although insurance is a low-interest product, if the data is not used in an ethical way, the use of algorithms could come under fire and consumer confidence could start to crumble. It is therefore essential to make sure that consumers do not lose their confidence in the insurance sector.
Professional development of algorithms
A vital precondition in this respect is that the technology is developed in a professional manner. There are obvious parallels with the rise of the internet. In the 1990s, every company wanted to get a website out as quickly as possible, and that job was often assigned to the person in the office who was 'good with computers'. A couple of decades later, the design and execution of online strategies has become highly professionalised. We see a similar process when it comes to the development of algorithms.
A lot of organisations make an enthusiastic start, only to find out the hard way that developing algorithms is another one of those tasks you can’t just give to somebody who is 'good with computers'. You need knowledgeable professionals. The difficult part, namely, is being able to analyse the data contextually, and not just statistically, in order to prevent any nonsensical conclusions being drawn. An essential part of data analysis, therefore, is close collaboration between application specialists in the business and data scientists. This process of professionalisation is a key building block in the establishment of trust.
More algorithms in the insurance sector
Would you like to know more about this subject? You can read all about it in our eBook - Algorithms in the Insurance Sector.