By Harry Rosenthal, General Risk Manager
We live in a world afloat with data, in which almost every item of technology leaves a wake of gathered information about us. Most people are aware of the data produced by personal fitness devices and smart phones tracking our movements, browsers collecting our search habits, and vendors tracking our spending – but few are aware that a single Boeing 787 flight typically generates half a terabyte of data. It has been estimated that, in the near future, internet-connected cars will each send 25 gigabytes of data to the cloud every hour as they drive us around.
An interesting phenomenon of this century has been our willingness to give such personal data away for very little in return. In exchange for constantly giving our location to phone providers, we receive “Find Your Phone” apps. For the pleasure of free online searching, we give Google all our search and online shopping histories. And for free access to social media platforms, we are willing to share a great deal about every aspect of ourselves and our networks. These seem like small returns for sacrificing our valuable privacy – but how valuable actually is that privacy?
Unsurprisingly, risk financers such as insurance companies have been early adopters of mining big data and of the artificial intelligence (AI) that gives it meaning. Much has been written in the media about insurers starting to grasp the value of large data sets, and many are investing in the artificial intelligence needed to make those data hoards useful for understanding risk. For example, QBE is a leader in this area through investing in AI companies and developing machine-based learning platforms for analysing policy wordings.
Insurers say that there are significant potential benefits for both customer and underwriter from using AI and big data to better understand risk. Such insights could lead to improved policy wordings and more accurate premium assessments for risks, as insurers will have multiple sources of data from which to ascertain a risk, and won’t need to rely solely on what the applicant reports.
So, what will this future look like? In the next few years, insurers will develop better computing abilities and AI capabilities to improve risk modelling. At first they will apply these tools to their own data sets, such as underwriting applications and claims experience, but it is likely that others with large data sources will soon avail themselves of such tools, whether through direct purchase or by investing in technology. For example, a New Zealand company, Rocket Lab, recently put three commercial satellites into orbit via rocket launch – a service it expects to cost just US$4.9 million per flight. Such cheap launch technology, when mixed with the recent development of low orbit nano-satellites, will allow insurers to gather geographic and surveillance data at reasonable costs. It is not much of a leap to then envision insurers tracking our smart cars via their own satellite networks in order to better understand policyholder driving habits.
What about keeping “risky” clients under observation? For example, such nano-satellites – no larger than a mobile phone – could also be used by insurers to keep an eye on developments in flood-prone regions. It is beneficial for local councils to develop areas that are attractive to high-end home buyers, but these areas – which often include waterfront areas, river view lots, or expansions into flood zones – can be risky to insurers. While the council might promise mitigations such as improved drainage or levees, such works don’t always keep pace with the lucrative developments of coastal and other waterside communities. With inexpensive space-based surveillance technology – and the AI to capitalise on the data streams it generates – insurers will be equipped to change the way in which products are sold and underwritten.
There looms, therefore, the very real possibility that the insurance market as we know it will experience significant disruption as a result of this profusion of personal data meeting advanced AI. While insurers continue to explore the help of these new resources in improving underwriting and the accuracy of forecasting models, such technology, once loose, tends to be adopted by all parties, not just its inventors.
From the insurers’ perspective, they are likely anticipating that consumers will consider providing personal data – such as fitness tracking from a device – as an advantage if leads to better coverage at a lower price. Other relevant information might include applicants’ smart home readings, smart car printouts and even smartphone records. In exchange for relinquishing this privacy, consumers could receive substantial premium savings.
Experience teaches us that technological advances can cause interesting disruptions. New technology sometimes goes in unanticipated directions – just ask taxi drivers, small hotel operators or bookstore owners. Each of these have been disrupted by new computing software. If we acknowledge this sea of public and personal data as the momentous resource that insurers believe it to be, then other parties will be quick to follow in recognising its value and using it to change traditional customer/client relationships, including, perhaps, the traditional insurance “applicant & underwriter” model.
What I envision developing out of this confluence of factors is the creation of something I’ll call the Insurance Consumer Exchange. This would be an online, data-driven customer pool of insureds. Rather than seeking new customers through advertising, insurers will have to adjust to a new acquisition mechanism in which potential insureds control the process. In this independent market, insureds will willingly provide large amounts of raw personal data, to which insurers can apply their underwriting algorithms and AI to identify superior risks and pricing. A new class of client, without the help of intermediaries, will emerge – marketing themselves through the offering of their identified, personal data and waiting for interested global insurers to bid for their business.
Typical data sets could include real-time results from personal fitness devices, records of gym membership and number of annual doctor’s visits. Perhaps the computers in our cars will report on our driving characteristics, such as distance, speeds and driving times. Maybe our credit card companies will report our spending on alcohol, cigarettes, home maintenance materials, electricity bills, and so on. Like a reverse Gumtree, insurers will be able to access this exchange of individual data and review it when insureds’ current policies are close to expiration to provide a new quote. All interested insurers will be invited to offer terms and covers to their preferred prospective insured customers, and the provider that offers the best overall deal will be successful. This will be fully automated with no need for advertising, marketing staff, celebrity endorsements, umbrellas or golf balls to attract new business – instead, computers and AI will replace all most all existing customer acquisition costs.
The catalyst behind this sea-change will be the recognition, by owners of large amounts of data, of its value and utility. The data we leave in our wake will no longer be free to all, but rather will be consensually exchanged for something of tangible value, like lower premiums and better policy wording. Each insurer’s AI network could make an automated offer for our individual life, health, home and contents, auto and travel policies. One can assume that insurer’s computers will know a good risk when they ‘see’ one. I would expect all insurers to offer the most competitive quote in order to obtain their chosen customers. The consumer would have a passive role, only accepting the most attractive bids as based on their own AI algorithms.
It’s possible that the future won’t look quite like this, but what is certain is that, over the next few years, data owners will start to recognise the gold in their printouts and try to convert it into something beneficial. The days of freely giving up data for small favours are coming to an end – and the use and abuse of data will be one of the defining characteristics of the rest of this century.
The recent Universities Australia conference discussed how, with a workforce increasingly exposed to the development of AI, the question is no longer ‘what do you want to be when you grow up’, but ‘what will you be when the robots grow up?’. To learn more about the potential impact of AI on society and other topics covered at the conference, click here.