The journey to Artificial Intelligence in Insurance

There have been a bunch of crazy developments in the field of Artificial Intelligence (AI) in the past few years. Google’s AlphaZero beat the leading Go player in 2016 using techniques called Neural Networks and who knows what they’ve achieved in the 5 years since. If you’re interested they publish most of it here: https://deepmind.com/

This blog isn’t really about AI though, it’s about the journey that companies need to go on in order to achieve AI-readiness.

What is AI?

Artificial intelligence (AI) is intelligence demonstrated by computer programmes that mimics human-style intelligence such as learning and problem-solving, as opposed to more traditional ‘computer’ skills of calculation or processing.

Why is AI relevant for insurance brokers?

AI has wide relevance for almost all spheres of industry, but insurance particularly because of the vast amount of data and variables available to analyse. Healthcare has seen a huge surge in AI investment in recent years and it is inevitable that some of this will trickle down to insurance products, journeys and pricing too.

What are the stages to get there?

Most businesses are a long, long way off achieving AI. But there are steps along the journey that they can be making.

Data Access

AI feeds off data. Access to data is therefore mission critical the beginnings of any AI venture. Companies should ensure that their data is readily accessible, structured, and as full as possible to make the most of it, now or in future.

Data Analytics

Humans are good at problem-solving. Better than most machines. Once data has been assembled there are good tools (like PowerBI or QlickView) for visualising it. Businesses should layer such tools onto their data warehouses/lakes and make the effort to analyse the results of just looking at trends and correlations.

Machine Learning

Machine Learning (ML) is the step between looking at an Excel spreadsheet and AI. ML is not ‘intelligent’ in that it needs to be told ‘what’ to do, but it can do what it is told to do better than a human can. ML, properly trained, will notice correlations within data that the human eye cannot spot: it will see multi-factor or multi-dimensional correlations for which there are no easy visualisation tools. Microsoft Machine Learning provides an excellent toolkit for such projects. Building supervised learning algorithms will likely require an experienced or trained data analyst or programmer.

Artificial Intelligence

Once these initial steps have been taken, the door to real AI is open.

Very few insurance companies truly stand at this hallowed portal as yet, but for those that aspire to do so, the previous steps must all be taken. And there is great benefit to be had for the business and its’ customers in doing so.