What Will Autonomous Cars Do to the Insurance Companies?



What Will Autonomous Cars Do to the Insurance Companies?

Authors

Iva Bojic, Massachusetts Institute of Technology, Roman, Braendli, Carlo, Ratti

Description

Insurance companies are in the business of assessing, pricing and mitigating risk. In complete concurrent information equality of individual risks in transportation, which autonomous cars will bring, historic models lose their relevance.

Abstract

Insurance companies regard themselves as being in the business of assessing, pricing and mitigating risk. Customers accept a certain loss (premium paid) for the guarantee that in case of an uncertain future event, the insurance company will compensate for that loss. Historically, insurance companies have based their business models on two competitive advantages: (i) risk assessment, the superior knowledge of historic causal event data for loss probability, and (ii) risk pooling, the inability or unwillingness of entities to cover their total individual risk. In the case of car insurance, historic population risk data (e.g. an individual risk profile’s probability for car accidents), and inability or unwillingness of their customers to take their own risks (e.g. paying on their own for the costs of repairing or replacing their car after an accident) compose said advantages. Today, a future paradigm shift around car mobility becomes evident. In the face of new technology, behaviors and business models, can historic competitive advantages be sustained?
Advances in transportation solutions will certainly cause a disruption in the lifestyle we know today and will most definitely contribute to information equality. In complete concurrent information equality of individual risks (i.e. companies and their customers sharing the same access to information), historic models lose their relevance. Even now, companies like Uber or Google gather more information on drivers and the mobility ecosystem than incumbent insurance companies do. These data sets allow individual and predictive assessments far beyond what historic models and diffuse customer profiles offer. Unchanged, insurance companies will lose the competitive advantage of risk assessment.
In the approaching era of autonomous cars, human error, which today accounts for more than 90% of causes of car accidents, will be all but eradicated, significantly reducing the overall risk available to pool. Building on these predictive models, we may soon reach a pivotal point where human drivers are assessed as disproportionate risk factor in a largely machine-driven mobility environment and pooling of risk for them rendered so individual and above total risk pool as to render it unappealing to customers. Pooling of risk will no longer be limited to an insurance portfolio or individual profile, but rather to a bigger data sample on city- or population scale, based on ubiquitously available population data.
All but eliminating individual risk is equal to trusting machines to make real-time, data-based decisions. Car-based mobility, however, can no longer be assessed ceteris paribus. Mobility is headed towards an integrated system and reducing risk factors will only be possible within that autonomous system. With autonomous vehicles programmed to kill, how will machines assess the risk of ending an individual’s contribution to the city or society? Faced, for example, with the choice of killing a driver or a pedestrian, autonomous vehicles will have to be programed to choose lesser evil. And as we are entering deeper and deeper in the digital age, machines are going to have more and more information on basically whose life worth less.

Publisher

Association for European Transport