For those of us who learnt to drive on Guernsey roads, it can be an interesting experience. What Guernsey lacks in opportunity to drive at higher speeds and on motorways, it more than makes up for in the challenges of having to avoid pedestrians, tractors, cyclists, horses, continual pavement mounting, other vehicles and of course the dreaded granite walls!
There is no opportunity to let the mind drift on Guernsey roads, it requires attention and continual engagement to avoid any of the aforementioned pitfalls.
So it is always interesting to hear how the world is apparently going to be overtaken by driverless cars. As the seemingly unending number of small granite scrapes and near misses attest to in the islands, it makes one wonder how intelligent these driverless cars will need to be in order to navigate our haphazard roads.
Guernsey Roads are for the most part a beautiful part of our island. We are surrounded by aged granite walls, covered in lichen and the road conditions (usually as a result of them having been dug up so much!) for the most part are excellent. But they do follow the natural contours of land ownership and traditional thoroughfares. As such, the roads are accommodated within the existing footprint and so there are lots of bends and turns, lots of hidden entrances, which is the way it has to be when you have such a plethora of traditional buildings throughout the island.
So AI driverless systems that were designed for the open roads of America and the broad highways, where turn-offs are infrequent and speed limits are higher, are not suitable for the smaller rural lanes of Guernsey. And it’s not just Guernsey. Some rural parts of the UK and Europe share the same characteristics as Channel Island roads. They also have to accommodate ancient buildings and in some places are simply too narrow for two cars to pass abreast of each other.
Car insurance providers in Guernsey understand this. We know that over time drivers develop a unique sense of how to drive on the island and in many cases they are experienced in taking specific routes on a daily basis. Any of us crawling our way towards the Grange, or along the seafront to try and nab a ten hour parking spot will know of that pain! Also what about diversions? Does a computer have the capacity to know to take an alternative route or in some cases to actually understand the sign writing on those diversion signs (UFN anyone!).
But what is unknown is how robots as opposed to humans will react to Guernsey roads. If these safety tests have been conducted on motorways and large dual carriageway both in the USA and UK, does that mean we will have to have separate safety sign-off for the Channel Islands? Would anyone (politician or official) be brave enough to sign-off on automated vehicles, knowing that the first crash that happens is going to result in some degree of blame and questions over the safety of such vehicles.
Also, will automated vehicles understand the nuances of Channel Islands roads. That means, where they know they can mount the pavement safely, will they do so? And will the vehicle come to know that having turned the very tight corner and avoided a granite wall, you are likely to clip the corner of the pavement forcing the car to lurch slightly? These are everyday considerations that Channel Islands drivers have to contend with and have experience to deal with. But will a computer take an overly cautious approach and then fundamentally impede traffic and cause unnecessary delays, or potentially worse, take a more risk tolerant approach and thereby cause accidents.
It is also the conundrum that no one has been able to answer just yet: say a schoolchild is walking along the pavement to your left and a vehicle coming at speed towards you from the other direction. Does the computer decide to plough on into the oncoming vehicle and rely on the vehicle safety measures to protect passengers. The conundrum is then one life vs. multiple passengers. And if it decides to sacrifice one child to save the larger number of passengers will it then decide to mount the kerb, supposedly trying to save the multiple occupants of both vehicles by hitting the child instead. Both are totally unpalatable. At least with the human they may spot something, see some avenue to avoid both. With the computer, this is a simple case of proceeding forward and does the algorithm decide who is least valuable? A designer will say this is a remote risk, but the whole thing about remote risks is that they still have to be accommodated for. And would you necessarily trust the computer to make that decision over a human? Some would, some wouldn’t. Either way it could potentially be a tragic outcome.
Some might say this is a focus on the negative unnecessarily. Yet as car insurance providers in the islands, these are questions we have to grapple with. As a small community we also have to take a slightly different approach from a risk perspective than other jurisdictions. What might be good for them might not necessarily be good for us.
This is also not to scare nor discourage the wonderful innovations that such technology can bring. Our childhood cartoons had us believing that we would be flying to work, and yet as I fill up my 900 cc Honda van at Doyle’s, I can’t help but think how wide the gap is between expectation and reality. And that’s not always a bad thing. Where we can benefit from technology, we should do so. But unfortunately the boring concepts of risk management must be accommodated. As much as we might get excited about the potential of some of these developments we also have to temper that with the understanding that Guernsey is still a unique place and that we therefore may need to take a different approach to the world at large. Predictability is for the most part safer than the risks associated with the new. Maybe we simply allow others to trail it exhaustively elsewhere first and then take a decision. At that point we may then also be able to copy and paste the supporting legislation needed, once it has been tested in other jurisdiction’s courts.
If you want my prediction, and this comes with the caveat that it is just that, I think we’re still a good 20 or more years from seeing any potential demise in people driving themselves. This arises from the desire to continue driving, the mistrust of technology and not having technology which is capable of understanding quirky street design. I think therefore that the rollout would largely be in the cities and will try to accommodate specific trades such as courier and taxi work increasingly and so private car technology will remain the last hold out. That said, as more city-dwellers decide to forgo owning a vehicle themselves, it is likely that rental fleets will proliferate with driverless vehicles in the coming years. There will also have to be an improvement in the technology which supports the infrastructure. Better use of on-board radar, which will have to reach an incredible level of sophistication. This will also then tie in with a form of Google Maps that effectively scans and produces a 3D layout of the roads and buildings and other perils that will then assist the on-board computer to navigate around for instance, blind corners.
If you really want to engage in some left field future thinking, imagine then a world where everyone else has automated vehicles but the Channel Islands and others have still not managed to accommodate them. At that point, to manufacture self drive cars for humans and for such a limited market would be uneconomical. At that point we may have to take our chances. I think by the time that comes around though, I will probably no longer be driving anyway for one reason or another!
The only major advantage and speaking as a local insurance broker – wing mirrors – in theory a driverless car would not need them. Given we have to pay to replace hundreds of wing mirrors a year – perhaps I should be more supportive of automation!