The spat between Uber and California regulators over permissions for the ride-hailing company’s autonomous vehicle fleet has exposed the degree to which information technologists believe that the world can be treated as a Petri dish for their experimentation. Innovation comes first, on this view. Unintended side- effects can be dealt with later.
In this case, Uber objected to being asked to enrol in the California state regulator’s specially designed autonomous testing programme on the grounds that rivals such as Tesla had been able to bypass the programme by uploading autonomous tech directly to its network of vehicles. Uber also took exception to the wording of the Californian regulation, noting the rules apply to cars driven “without . . . active physical control or monitoring”, while its vehicles were still incapable of this. The Department of Motor Vehicles disagreed and revoked Uber’s registrations.
Disgruntled, Uber packed up its fleet and headed off to a more accommodating regulatory climate in Arizona. This response reflected a contempt for a regulatory process that aims only to weigh up the unseen costs of innovation against as yet unproven benefits.
Regulators must be convinced not only that autonomous cars can operate without incident but also that they have data showing how many times human operators were forced to intervene and in what circumstances – data the public would otherwise not be provided with. In addition, they must understand how autonomous cars are perceived by other drivers – the impact they have on productivity, wider driving behaviour and traffic, not to mention how vulnerable these vehicles are to software crashes or hacking.
Tech companies understandably fear that too much data sharing could result in public officials overreacting in the event of innocuous setbacks. But this is a poor justification for non-co- operation. A useful comparison can be made with the pharmaceutical industry. It, too, is a highly innovative sector. Yet there are regulatory constraints on how its innovations are tested and rolled out.
Statistical feedback is a critical consideration throughout the development process, as are cost factors and, most important of all, the scale of potentially harmful side-effects. And when considering negative side-effects, regulators also take into account economic and psychological impacts. For many new drugs, if the costs outweigh the benefits, they do not make the grade.
Yet, for some reason, despite its significant impact on society and culture, attitudes to the deployment of IT have been somewhat laissez faire. Tech companies roll out devices – software, virtual reality, you name it – with scant consideration of their longer-term impacts on society.
Only recently, for example, have sociologists and psychologists recognised that smartphones could be having detrimental effects on cognition and attention spans. Some surveys even suggest smartphones may be making us less productive.
It is interesting, in this connection, to consider the attitude of the traditionalist Christian Amish community in the US to new technology. The Amish are often depicted as Luddites. But this is misleading. They are, in fact, selective and thoughtful about the technology they choose to adopt. Mindful of the potential effects on their distinctive culture, the Amish prefer to wait and see how technology develops elsewhere. If the effects prove to be positive, they are open-minded about its adoption.
Going the way of the Amish is clearly not an option for a highly developed modern society. But it does seem reasonable to ask technologists to show, through rigorous and transparent public testing, that they are really acting in the public interest.
