Driverless cars shrouded in misplaced assumptions

As the discussion over driverless cars heats up, like any contentious debate, the arguments for and against become fuzzier and more readily based on assumptions.

I recently wrote an opinion piece saying that one overlooked issue is my assumption that most, if not all, autonomous cars will be electric. This may appear to have been a cavalier statement, but my thinking is that any car that’s so advanced as to be self-driving, would be utilizing the latest technology, including an electric rather than gas-powered powertrain.

But in the budget environment, funds simply don’t exist to develop the roads and highways infrastructure required by electric vehicles, driverless or otherwise. And while electric vehicles represent the way things seem to be going in the UK, in our much vaster country, it’s hard to imagine this kind of investment being made in the near future.

…Back to autonomous cars

Popular opinion would appear to be in support of autonomous cars. Some argue that driverless cars making roads safer in the UK, where they’ve begun testing driverless delivery vehicles in earnest. Artificial intelligence learning is cited as the key factor in favor of the “safer” argument. And all the numbers stack up to this.

However, we live in a country, a society, of human beings, not numbers. Our fears are often irrational, and legion. Our leadership and media do everything they can to stoke and sensationalize those fears.

But at a more practical and pragmatic level, we have the insurance industry, which is all about classifying risk, pricing it, and to the extent possible, mitigating it by telling us what we can and can’t do, under a given policy. This is our world, and what I can say, without getting into murkier or more philosophical territory, are the unassailable facts of our land, our road system, and our insurance industry as they relate to driverless cars. And based on these realities, I still submit that it’s going to be quite a while before autonomous vehicles are viably, numerously and yes, safely on our roads.

Related: Driverless cars? not so fast

Consider the actual roads

There are 2.7 million paved and 1.4 million unpaved roads in the U.S. Would you have guessed that there are 50% as many dirt and gravel roads as paved ones? And three-quarters of them are located in rural areas.

What does this have to do with driverless cars? One would think that traversing all those country miles would be a great use for driverless vehicles — especially delivery vehicles.

The trouble is that robo-cars can’t be driven on roads without lines painted on them. And while there is likely a sizeable armada of technicians working on this problem, there has yet to be any viable solution to describing, in computer language, what autonomous vehicles need to do in these circumstances.

It’s an issue of asking the right questions, which are so numerous that no one has even begun to crack the code on it.

Related: America’s roads are a drag on U.S. auto insurers, consumers

Technological shortcomings

There are major gaps in what computers are capable of, especially in self-driving cars. Let’s start with the fact that these vehicles seem unable to function well on different terrains and in inclement weather. Occupant safety depends on the breadth of vision of the programmers who are working on these issues.

We if we were to become a nation of coders, rather than a nation of drivers, we’d still be a long way from filling this gap.

The impracticality of driverless cars also includes that they won’t go down gravel roads, they won’t do well in snow and ice, and they won’t do well in varied light conditions; these are only some of the situations the programmers in Silicon Valley have yet to figure out.

So we have, currently, driverless vehicles in the test phase that can’t be used on the vast majority of the roads in America. They won’t be able to be used on these roads until someone figures out how to resolve the numerous gray areas inherent in transportation infrastructure that was built by and for people, and is therefore flawed, inconsistent, and in a constant state of change.

Then there’s the uninsurability issue

This is something I’m closer to, as my own work focuses on bringing about automation of processes in the insurance industry. Cars have been on the “compulsory insurance” list since 1939. But it’s really the drivers, not the cars, who are insured.

Now, with the driver taken out of the equation, the insurance implications change completely. In particular, in the determination of fault for a tragedy such as Uber’s recent accident in Arizona, the typical insurance framework we have today will not work with driverless cars.

Though this is by no means the first time a radically new product has appeared and people didn’t know how to handle the risk, gross and comparative negligence determinations are predicated on the idea that there’s someone in the driver’s seat, literally. In the pre-Model T days of the automobile, there were laws on the books that said if you were driving a horseless carriage, you had to have somebody on horseback in front of the vehicle warning people that a car was coming. To serve as a warning to others, cars would have to be preceded with a lighted lamp. As cars proliferated on the streets and accidents started happening, Travelers Insurance was born.

With driverless cars, driver liability goes out the window. The insurance becomes something much closer to product liability insurance, because the fault of the accident, presumably, was with the equipment. And the very small number of driverless cars that are on the road — compared with tens of millions of existing cars that are not driverless — means that, for the present, this is a quirky anomaly that almost needs to be handled on a case-by-case basis. In essence, they’re putting these cars on the road before they’ve figured out the insurance realities.

So I once again, I rest my case: Driverless cars are not ready for prime time in America.

John Sarich is an insurance industry analyst and VP of Strategy at VUE Software. He is a senior solutions architect, strategic consultant and business advisor with over 25 years of insurance industry experience. He can be reached at John.Sarich@VUE Software.com.

These opinions are the author’s own.

See also:

Driverless cars may cut traffic jams, not insurance premiums

The shift to autonomous vehicles means opportunity for insurers