Out of our hands

This is my editorial for the Winter 2016 issue of HardCopy magazine:

The world as seen through the sensors of a Google self-drive car.

In February 2009, in an effort to demonstrate exactly how far we are willing to obey a seductive voice emanating from a plastic box, the driver of a 50-foot articulated lorry wedged his vehicle so thoroughly into a hair-pin bend in the tiny Cotswold village of Syde that it took five days to extricate. In the light of this and other such examples, I am heartened by the news that Google, Amazon, Facebook, Microsoft and IBM have announced the Partnership on Artificial Intelligence to Benefit People and Society in order to look into the ethical and societal implications of such technologies.

Many of these implications stem from the lack of human involvement in the decisions that such technologies are increasingly making. Driverless cars are almost upon us, and by most accounts orders of magnitude safer than human drivers, particularly in cities where they can communicate with each other to better understand the dangers ahead. However, on 7 May this year a Model S Tesla in driverless mode smashed at high speed into an 18-wheel truck and trailer, killing the ‘driver’ instantly. It looks as though the car was unable to distinguish the white truck and trailer against the bright Florida sky behind, something that most human drivers would be able to do ‘without thinking’. It is this that
makes such accidents seem ‘inhuman’.

Or there’s the customer who rings up wondering why the insurance premium you quoted him is twice that of his neighbour, whose circumstances appear pretty much identical. You have no idea because you have no way of understanding how the algorithm that your company uses arrived at that decision, and neither does anyone else. Does the customer have a right to be told? In other words, does the software have to provide some sort of audit trail? Maybe, but in the meantime, companies are understandably reluctant to divulge their use of such things.

AI is already with us and, as our cover features make clear, already capable of extraordinary feats. However it’s only too easy to anthropomorphise. It is conceivable that, some time in the not too distant future, some thing that could be called ‘conscious awareness’ may prove to be the most logical way for such machines to manage and comprehend the world in which they find themselves. However that is a long way from the establishment of a conscience, whatever that may be, and even then it seems unlikely that an intelligence made of silicon and metal will have any understanding of emotions or feelings, or anything resembling empathy for the strange organic creatures whose lives they will, by then, dominate.

While there is an obvious agenda to promote their shiny new services, the coming together of these companies at such an early stage does at least suggest they accept a degree of responsibility.