Today’s insurers operate in a world where machines are responsible for making operational and sometimes even strategic decisions.
Not only do we use algorithms to drive relatively easy decisions such as who to market to or which claims to investigate for fraud, but we are also increasingly relying on machines to make more complicated decisions like whether to write off a car or to repair it after an accident.
Artificial intelligence
This is not surprising, as machines and algorithms are generally more consistent, objective, and swift in making these decisions, and, best of all, are also far less likely to take a duvet day.
In this quest to develop systems that border on artificial intelligence and that can make increasingly more complex decisions faster and more accurately, it requires us to think differently, almost primitively, about the interaction of systems, data and predictive analytics in creating value adding solutions.
You can think of human beings (or at least a large portion of them) as intelligent systems where data is generated from our senses and then sent to our brains to be analysed. The brain uses these sensory data inputs to achieve two main goals. Firstly, to draw a conclusion from the situation and send immediate responses back to the system to react to the input, and secondly, to learn from the experience so that the situation is dealt with more intelligently in the future (i.e. if the stove is glowing red – do not touch it, it is hot).
Improvement over the years
With the rise of predictive analytics, machine learning, actuarial science and big data in the business world, we have successfully replicated and substantially improved on these two functions over the last couple of years.
This is particularly evident in operations like insurance where a large number of decisions need to be made with a lot of data inputs in a very short space of time. Players in the insurance arena are at the forefront of developing really innovative systems and algorithms. However, if we really want to get to a point where we see the true value that artificial intelligence can offer, isolated innovation just won’t be enough anymore.
Real break-through
One critical element that defines human intelligence is our ability to rapidly accumulate and access knowledge without having to experience everything ourselves. It starts with something as basic as passing on a story over generations or a mother telling her two year old not to touch the hot stove.
Historically, though, we saw the real break-throughs in the evolution of human intelligence when new technology enabled this knowledge-sharing to happen on a large scale. Think of the invention of paper, the printing press, the telephone, the personal computer, the internet and smartphones, and the impact these technological advances had on how people shared, and continue to share ideas, learn and make decisions.
In some regards, we have learned from this in the business world. We want intelligence in our businesses to evolve at a faster pace and short circuit the learning-through-own-experience process. We want to piggyback from other companies’ experiences without having to step into the pitfalls ourselves. However, the machines making the day-to-day decisions in our operations cannot easily train themselves with this type of information. They need things like shared data repositories, API frameworks and cross-industry accepted standards on things like spatial data layers, asset identification and location.
True artificial intelligence
Focused innovation on these machine friendly knowledge sharing initiatives will enable predictive analytics and actuarial science in insurance to transition into true artificial intelligence in the years to come. This will not only unlock a tremendous amount of value for all industry players, but will fundamentally change the way the industry works by ultimately helping the end consumer manage their personal and asset risk better.