The AXA.COM team editor

Has "big data" become "too big"?`

The retrieval, storage and analysis of personal data have witnessed extraordinary advances in the last five years. Future of insurance
Dec 1, 2015

The retrieval, storage and analysis of personal data have witnessed extraordinary advances in the last five years. This trend is nothing short of a revolution which goes beyond mere technology and is causing far-reaching changes to our habits and lifestyles. That is why this amazing source of innovation is also raising concerns, particularly when it comes to privacy.
Has "big data" become "too big"?`
Here we shed light on the key issues and identify areas for future action which are somewhat removed from conventional wisdom.

3 questions for Anthony Goldbloom and Eric Lebigot

Big data, a source of innovation ... and concern?

The overwhelming majority of services which we now use every day function thanks to data. What changes is this model driving? Why has it become predominant? And should we be wary of a service that we help feed into and improve?
The intersecting views of two experts are shown below. For Anthony Goldbloom, CEO of Kaggle, a company that organises data analysis competitions open to all, big data are about to revolutionise all sectors of the economy. Eric Lebigot, AXA Group's Chief Data Scientist, sees an opportunity to offer more personalised services while ensuring respect for privacy. Three questions to gain a clearer insight.

Why has there been a surge in personal data? What explains the big data phenomenon?

Eric Lebigot: The production of personal data is not new, but the quantity being generated has skyrocketed with the advent of digital technologies, the democratization of the Internet and the emergence of new online services through which people share more and more data. Due to these many new uses and interactions, a massive amount of data is created every second and can now be utilized, thanks to the acceleration of technology and the techniques it has made possible, such as machine learning.

The Big Data phenomenon is defined by “the three Vs”: volume, velocity and variety, which create new opportunities and challenges for companies (in particular for accessing, storing and analyzing this data).

Anthony Goldbloom: In addition, we are now able to store these data and so establish ever bigger bases. The technique for analysing them has progressed considerably. The data are more numerous, better stored and better analysed. This is the origin of the big data phenomenon.

Why are data now so valuable?

EL: Today’s data is valuable because it holds the keys to untapped opportunities and provides a genuine competitive advantage. What’s more, this wealth of information is increasingly being enriched with data available to everyone (e.g. weather information) and data exchanged by various companies. This enriched data, combined with today’s powerful machines and predictive techniques, now enables companies to refine their understanding of customer needs for the development of new products and services.

An added benefit for AXA, as an insurer, is that a sophisticated analysis of data gives us a better vision of the risks our customers face. It enables us to be even more accurate and precise in assessing our policyholders’ risks, to provide competitive quotes, or in evaluating complex situations, such as exposure to natural disasters, to enhance risk prevention.

AG: In fact, data are a source of innovation. Today, data scientists (or data analysis experts) can find solutions to health issues, for example, even before healthcare professionals. I see this as one of the key aspects of the big data revolution: it is fundamentally meritocratic. The best solution wins, not the best diploma or highest brand awareness. All those who feel they are capable can attempt to respond to the problem posed.

Should we be wary of the use made of these data?

EL: Everyone should be aware of how their data may be used, so they can choose which personal information they want to share, and for what purpose, and perhaps take steps to restrict the data they provide. Still, refusing to share certain types of data sometimes means that all services are not available. I think that a good model would be a win-win use of data, where certain services are provided in exchange for a little personal data: data which often enable access to enhanced or new services.

As with every major new innovation, the regulatory framework must adapt to this changing environment. The new European Global Data Protection Regulation, which will enter into force in early 2018, will also significantly bolster citizen’s rights in terms of personal data privacy, while increasing the powers of data protection authorities by introducing dissuasive sanctions for companies.

At AXA, we are strongly in favor of making a firm commitment to ethical use and protection of data, because we believe this aligns us with our customers’ values and builds their trust.

AG: The use of data will transform all sectors of activity without exception. And, like all revolutions, it will create injustices, uncertainties and progress. I believe that we are at the end of the first phase, where technologies have reached a certain maturity. Today, we are reaching another stage where these advances have to be integrated into the daily lives of users without causing rejection. This is a crucial stage and one that I believe is much trickier to negotiate.


Eric Lebigot spent 17 years doing research in theoretical physics (quantum physics and gravitational waves) in the US, France and China, working in parallel as a scientific consultant for the industry, and also more recently co-founding a start-up, IviDATA Tools, where he became Lead Data Scientist. Eric joined AXA in September 2015 as Chief Data Scientist AT the Data Innovation Lab.

In 2010, Anthony Goldbloom launched Kaggle. This web platform organises data science competitions. Companies make their data available to data scientists in order to solve a problem. "Kaggle is a very simple way to answer complex questions," explains Anthony Goldbloom.


Big data at the service of policyholders

The surge of information originating from big data poses new challenges to society in terms of the ethical boundaries to be fixed for their use by companies or public organisations. Insurers have a role to play in limiting cyber risks and developing tools to exploit big data in the common interest.

Big data remains somewhat mysterious to the general public and yet they are opening up new and fruitful perspectives in a multitude of economic, biological and social fields. The key is to retain a sense of adjudication in order to use them wisely.

Daniel Kaplan, co-founder and CEO of the Next-Generation Internet Foundation (Fing), a think tank which defends the transformative potential of new technologies, places considerable hope in a society which will create the means to collect personal and impersonal information on a massive scale without losing its free will. Although the general framework for data governance must be defined by states, insurers also have a role to play in order to make big data beneficial to individuals and organisations while minimising excess and misuse.

Insurance as a data aggregator

François Ewald
Professor at the CNAM (French national centre of arts and trades)

The data revolution is creating several challenges for insurers. A technological and industrial challenge but also a societal one. Insurers will have to upgrade their organisations and techniques so that they can use big data to provide better services to policyholders while offering genuine added social value. The insurance industry must assume its role as a data aggregator. It is about switching from the concept of risk as an event to that of risk in the broader sense.

Although insurers moved beyond their traditional claims management function a long time ago to place themselves upstream of risks, the surge of big data offers them a technological opportunity to further strengthen their role in terms of prevention. Big data information can indeed be of great service to society as shown in the UN's Global Pulse initiative which aims to draw together data from the most diverse experts in order to better anticipate future food crises or climate change.

It is about switching from the concept of risk as an event to that of risk in the broader sense

Lend support before, during and after the loss

It is then up to insurers to use these indicators to develop their own predictive and analytical models and respond more effectively and rapidly to policyholders when it comes to calculating and paying compensation. Big data then constitute a valuable material for the security of goods and people to be passed on to the public and businesses. This is the objective of the My Online (Mein Leben-Online) campaign developed by AXA in Germany as part of its general information security policy.

Juan Herrera
Head of Advertising at the AXA Group

We have created a site where we offer tutorials, expert opinions and a whole host of practical tips to allow policyholders to better protect their privacy and that of their family on the Internet, whether on social networks, using applications on their mobile, or when shopping online. The idea is to communicate and educate the public about the multitude of personal data circulating on the Internet. AXA supports policyholders not only during and after the problem but also before it even happens.

According to Daniel Kaplan, CEO of Fing, big data means that insurers should soon be able to offer prevention services to businesses in order to increase well-being in the workplace.

Educate the public about the multitude of personal data circulating on the Internet

Create new tools

More generally, players in the insurance sector have an historic opportunity to play a safeguard role in respect of the growing use of information generated by big data. It is up to them to develop the tools to analyse and supervise its impact on society from an ethical and environmental point of view.

New responsibilities could take the form of algorithm audits and various partnerships with NGOs or universities to use the data for the common good. But whatever the technical level of these tools, it is ultimately the acknowledged expertise of insurers in risk management which should enable stakeholders to exploit the new opportunities offered by big data within socially acceptable limits.

Daniel Kaplan
CEO of Fing

Big data must not be seen as a pure emanation of the truth. Its use for commercial or social purposes must remain an ongoing subject of discussion and debate under the adjudication of humans, not algorithms, no matter how complex they may be.

Did you know?

Parametric insurance, a high-precision tool at the service of policyholders

First developed in the early 2000s, parametric insurance covers new economic risks related to climate disruption thanks to satellite images and the big data revolution. Insurance seen from the sky!

1. Beyond natural disasters

... There are also "simple" weather anomalies caused by global warming: a record-breaking dismal or abnormally dry summer, winter temperatures well above the seasonal average, etc. Though less visible, these climate disturbances have consequences which are potentially just as devastating for many sectors of the economy.

2. Satellites at the service of insurance experts

With a high level of precision, in the order of 250 m on average, with a resolution which should eventually touch on 50 cm, satellites are able to define (and anticipate) weather anomalies as closely as possible. And, above all, they enable improved (and faster!) compensation for victims.

3. Parametric insurance knows how to use big data

Satellite images provide numerous metadata (wave height, biomass growth, wind speed, etc.) which permit a more accurate assessment of weather-related risks. These data make it possible to anticipate the effects of climate change on the activities of policyholders and to warn them in advance.

4. From north to south, the same issues...

The parametric method is aimed at those sectors which are most vulnerable to climate variations: agricultural production in emerging countries but also economic activities which are dependent on the weather for so-called "mature" countries, for example, tourism, transport, construction and energy, etc.

5. Across the planet

Thanks to satellites and metadata, parametric insurance is able to offer solutions tailored to all policyholders worldwide. A potentially universal coverage of weather hazards which lies behind the partnership which AXA has established with the IFC-World Bank.

Big data is Changing Our Daily Lives

Is there a before and after when it comes to big data? What will data change in our daily lives? An inside look.