In my last blog I elaborated on the impact of Internet of Things (IoT) for the Insurer (https://www.capgemini.com/blog/insights-data-blog/2016/08/iot-insurance-of-things) . I argued that large volumes of data and new powerful analytics methodologies gave Insurers unprecedented insights thus enabling them to develop complete new offerings and services.

Now let’s look at another aspect of these new possibilities and their adoption in day to day business. Many predict that the Insurance industry is on the brink of dramatic change. In hardly any industry is the combination of low client satisfaction, low interest products, highly digitalized products and high volumes as apparent as in this industry. No wonder that the threat of new entrants and Fintech startups is as high as it is here. And Fintechs are being funded by investors that truly believe that this industry is about to change.

Just two days ago a Dutch Insurer announced the fact that they are going to invest in Big Data. That in itself is not something very noteworthy as most insurers use analytics and big data in their core processes now to better understand the client or e.g. combat fraudulent behavior. The thing that struck me in this announcement is the fact that this insurer took the decision to train a group of 11 employees from various departments of their organization by enrolling them in a course on an Academy specialized in Data Science. This is truly an example of an organization understanding the fact that Data Science is a multi-disciplinary science with a lot of Business angles and not “only” IT. And that brings me to the interesting fact of this blog; the human factor in Data Science.

I have been interested in the human factor of Data Sciences for a couple of years now. As the technological possibilities with IoT, Robotics, Big Data and Cognitive are rapidly evolving what about the human factor? How will we as customers, employees and of course society respond? Will we automatically embrace these new techniques and possibilities? Will we alter our “Business as Usual” based on (old) process thinking and adapt our processes automatically? Will society solve the ethical dilemmas in time? I don’t think so.

To really harvest the possibilities of powerful Data Science a number of Human factors should be taken into account.

The ability of humans in organizations to adapt.

An old saying goes “people are willing to change but do not want to be changed”. This is particularly the case in a change to an “information led organization”. I still see a lot of great initiatives initiated on the IT side struggle to get implemented on the business side (despite agile working teams) and vice versa. The problem here is not only resistance to change but also how the human mind functions. Kahneman  in his excellent book “thinking fast and slow” already explained us how the human mind works in dealing with new circumstances and the almost automatic way the mind works on holding on to old patterns while actually dealing with new. Imagine now that humans have to make decisions based on factual insights instead of a combination of experience in solving things based on old patterns and processes that are designed to these patterns. The old patterns will probably win.

The possibility of wrong insights out of data.

Data scientists are probably all aware of the difference between correlation and causality. In presentations on Big Data I often quote the excellent work of Tyler Vigen (http://www.tylervigen.com/spurious-correlations) on this topic where the favorite of the audience always is the correlation on drowning in a pool and the number of movies Nicolas Cage appears in.

This is a clear example of high correlation with no further meaning whatsoever and definitely no causality. Now imagine that less obvious flaws in the analysis leads to decisions based on totally wrong evidence and insights. I consider this a true risk of the trend where we enable analysis through powerful self-service BI tools on external data sets. Who is going to check the insights? And how are we going to check these? Should we than rely on gut feel and instinct alone? Or a combination and if yes how? These are big themes that have to be solved before entering into widespread usage of Self Service BI.

The prediction failure

One of my favorite quotes is by Twain “it is hard to make predictions especially on the future”. Although through the use of big datasets we are making progress in predicting the future we are still far off from predicting the future 100% accurate. There will always be new patterns that still have no base in the datasets simply for the fact that this combination has not occurred before. One of the more recent examples of this is e.g. the surprising outcome of the US presidential elections. After the event though a lot of people were able to explain based on evidence, trends, and new insights that were there but were not taken into account enough. Hindsight is still easier than looking forward.

Society, ethics and laws lagging behind to new possibilities or restraining them.

As in any field that is rapidly changing with huge impact on society we struggle to adapt to the new reality. Already a lot of things are possible in the field of e.g. client insights, behavioral analysis, fraud, etc. However due to privacy or other laws there is a severe impact on the possibility to deploy these new techniques and insights. It is not my place to start a full ethics dispute here but I am simply stating the fact that laws and ethics will always lag for good or bad reasons thus hampering fast deployment. A very interesting example of where technical possibilities interact with ethics is the self-driving car. How will we program a self-driving car when it comes to the dilemma of who to protect in a possible accident when the car approaches a pedestrian crossing? Should it protect the driver and crash in the family with children crossing or should it protect the family and swerve leading to crashing in a wall killing the driver? (And if you chose for saving the family would you feel comfortable entering that car?)

Conclusion

We life in exiting times when it comes to the new possibilities that we can deploy in almost every aspect of our society. Robotics, automation, cognitive techniques, Big Data and powerful Data Science are opening up an array of new business models, products and services. However the human side of handling and make truly good use of these possibilities are still very much dependent on human capabilities to adapt and apply. This calls for not only a focus on the technique but also a focus on the innovation process. This process should embrace technique as much as (human) change and the process to come from ideas to implemented applications fast.

For the Insurers out there; the ones to survive will be the ones that are able to innovate, adapt and deploy faster than the Fintechs. Human factors will play a decisive role in that.

Save