That particular apocalypse arises from the premise that driverless cars will reduce accidents. The figure of a 90% reduction is often bandied about, but as I've previously noted, this is as reliable as tobacco industry-funded research showing that cigarettes are good for you. It's frequently elided with the vaguely similar claim that 90% of road traffic accidents are caused by human error, which actually means nothing. No doubt 90% of future robot malfunctions will be traced to human error, just as most smartphone or computer failures can be today. I imagine 120 years ago evangelists for cars were suggesting that they would reduce road fatalities by eliminating the risk of horses bolting. The one reliable statistic in all this is that 90% of what you read or hear about driverless cars is cobblers, quite possibly including this post.
According to The Guardian business pages, "The boss of one big motor insurer confessed privately a few months ago that he was mystified as to why his supposedly long-term shareholders never asked him about the threat that driverless cars present to his industry". This may be because they're not idiots. The fundamental assumption behind the "threat" is that fewer accidents mean lower premiums, but this doesn't represent a problem for the insurance industry because fewer accidents means fewer claims. In other words, turnover may reduce but profit margins can be maintained. In fact, the insurance industry might even experience a period of increased margins if it captures the initial efficiency gains of autonomous systems - i.e. premiums may prove "sticky". But while the prophets of doom may well be wrong, insurance does offer a useful angle to think about some of the impacts of driverless cars.
One promise of the robo-chauffeur is that we can move away from sole ownership towards an on-demand model. This idea is already emergent in car-clubs like ZipCar and hire services like Uber, not to mention leasing deals that cover repairs and allow the car to be replaced at the end of the initial lease term. An on-demand model doesn't just commodify the car as a series of rides, it theoretically reduces the number of cars required per capita and increases their utilisation, which will shorten their effective lives. A reduction in the amortisation period, together with the addition of autonomous systems, is likely to result in more expensive cars, while the increased technical complexity of those systems and the heavier wear and tear on the vehicles (even allowing for smoother handling by your robot Parker) will push up repair costs.
An overnight switch to driverless cars is logistically impossible, so the most likely scenario is a gradual transition, i.e. parallel running of "dual mode" cars with older manual vehicles (it doesn't matter for the moment whether the two types share road space or are segregated, though that would have a bearing on accidents). Given the expectation of lower premiums for driverless cars, this logically suggests two classes of insurance premium. But this class distinction will relate to the car, not the driver. This, together with the growth of on-demand cars, will reinforce the idea that insurance should be associated with the vehicle rather than the person, while the higher servicing costs will encourage the idea that this insurance should be extended to cover no-fault repairs. This has already prompted some insurers to suggest that liability for a driverless car should lie with the manufacturer, not least because this removes any dispute over the contribution of the autonomous systems in the event of a claim.
In other words, the anticipated fall in driver premiums (i.e. third-party cover) may be offset by an increase in repair premiums, and that increase may be folded into the cost of manufacture as a mandatory lifetime premium (in the way that a warranty is factored into the sales price today). Car insurance could simply dissolve into the cost of the car and by extension the pay-as-you-drive service. This trend would be reinforced if the need for regular autonomous system upgrades (a mix of hardware, firmware and software) further reduces the effective life of a car because it's simpler to deploy a newer model than rip out and replace its embedded systems. Though the cost of cars may increase with the addition of autonomous systems, their commercial treatment may paradoxically become more like that of commodities that have seen steady declines in their real cost, such as computers.
A consequence of this shift in the insurance model may be that car repair businesses will prove to be more vulnerable to disruption than insurers, both because technical complexity will drive many low-end operations out of the market and because insurers could negotiate better prices with larger repairers through volume deals. The repair companies best placed to negotiate with the insurers, and to bring expertise and volume to bear on costs, will be the car manufacturers. In other words, a likely consequence of driverless cars will be the disapperance of independent garages. Given the tighter relationship between car-makers and insurers, it may also make sense for large manufacturers to go into the insurance business themselves, much as they've already entered the financing market, extending their current provision of service plans to comprehensive service, breakdown and accident cover. The automotive industry has been shifting from commodities to services for decades, so absorbing insurance broking is a logical next step after absorbing premiums into the ticket price.
Another straw in the wind is the development of user or usage-based insurance (UBI), which uses sensors and telematics to analyse driver behaviour and tailor an insurance premium accordingly. Behaviour may currently be little more than the number of miles driven (i.e. connecting the odometer to the Internet), but it can also be extended to analyse driving style (e.g. frequency and strength of braking) and to adjust expected behaviour based on the outside environment (e.g. speed relative to road type). This is another example of the behavioural surplus of surveillance capitalism and how it can be monetised as a prediction product. While the current focus is on insurers as buyers of that product, there are many other possible applications and therefore buyers, particularly if feedback can be given to a hands-free occupant ("turn off here to buy that thing you want").
Driverless cars introduce data capitalists to the motoring industry, which is why Google and Apple are at the forefront of developments. Despite the hype about its prototype cars, Google's likely strategy is to sell autonomous systems (think of them as "appliances", in tech-speak), together with their superior data assets, to traditional car manufacturers. Apple's history is of a manufacturer that expanded into data, but by the specific strategy of locking-in its customers through proprietary technology. While it may be able to develop an iCar, this will probably be a high-end luxury good but quite possibly inferior to what the car manufacturers can produce themselves through the combination of their own technology and buying in Google's, in which case the iCar may prove to be as niche as the iWatch.
Google's model is egalitarian insofar as it wants everybody to provide it with a behavioural surplus. This means that it has an interest not only in making driverless technology available to as many car manufacturers as possible, but also in lobbying governments to move towards making autonomous systems mandatory. This will be easier to do if more people forgo owning a car in favour of hire services, and that means shifting towards inclusive (car-specific) insurance. Apple's model is elitist insofar as it only wants the behavioural surplus of high-worth users. This means it has an interest in driverless cars being introduced through segregation, i.e. in central urban areas favoured by the rich, but it will also want to monopolise the behavioural surplus by excluding third-party services. To that end, folding third-party insurance into an all-inclusive lifetime warranty (i.e. bumping up the ticket price) makes commercial sense.
In either scenario the motor insurance industry could be absorbed into the car manufacuring industry, but it could just as easily revert to a specialist wholesale service, invisible to the wider public, that pools risk for all car-makers, driverless technology manufacturers and data providers. This would make a lot of sense given that a single failure in future autonomous systems has the potential to produce widespread chaos and consequently very large claims (think of the Toyota and VW scandals of recent years). The motor insurance industry isn't going to disappear because of driverless cars, but it is likely to change radically. However this plays out, the one thing that would probably disappear (though this may just be my bias talking) would be those annoying TV adverts. Huzzah for technology.
No comments:
Post a Comment