Guest blog: How machine learning can better equip insurers

From simple one-way analysis to generalised linear models (GLMs) and data enrichment, the insurance industry is still evolving. Machine learning is the next step. It’s already transforming major industries such as medicine and entertainment, as well as the judiciary system, and we can learn lessons here.

Working with artificial intelligence and machine learning technologies forces you to think and learn from what’s been done in other industries. You need to be able to extrapolate from one industry to another. Data extraction used to be impractical or prohibitively expensive but today, intelligent cognitive systems and cheaper storage make it much more available.

Is it coming after our jobs? No, in a nutshell. Machine learning doesn’t replace actuaries and statisticians: it simply becomes another tool to help them work more effectively.

It will help us make better use of the data we already have, using more variables and more complex algorithms. It will improve decision making and our understanding of customers by giving us more accurate predictive models. It will help us to better tailor products and add-ons that meet customers’ needs. In the telematics space, machine learning can help improve the calibration of telematics scores and make sense of all the data coming out of the complex internet of things (IoT).

These systems still need to be trained, just like people. You establish it with a body of knowledge, then as you use it more, it learns more. Probably the one difference between us and computers is that we can’t do it on the same scale, or as consistently.

Ultimately, the success or failure of predictive modelling projects will depend not on the technology, but on how well we can interpret their results relative to actual customer behaviour. This is where these machine based tools can enhance decision making. Giving another mechanism to help us make more accurate and better informed data driven decisions. It’s an advance in technology like the office suite was for computers way back when. It is a great opportunity for people in technical roles to embrace the technology as the next leap forward in predictive modelling.

This technology is about augmenting what humans do. Some people will fear the change and see it as a threat. Businesses need to make sure everyone, from insurance professionals to end customers, understand its role and support its use.

As with any technology, there are pros and cons. It’s quick and easy to use, producing accurate and unbiased results. But undeniably in some cases it is difficult to understand and interpret, and can carry a ‘black box’ regulatory stigma.

It’s only a matter of time before this technology reaches the insurance industry. With the right investment and stakeholder buy-in, we can stop trying to shoehorn everything into GLMs and embrace machine learning as an effective new tool.

For industries that embrace machine learning, the future will depend on how well they marry its predictive power with old-fashioned human wisdom.

Jason Cabral Bio