The advent of ‘driverless cars’ has been regularly announced for some time. They refer to one type of ‘autonomous vehicle’ (AV) tested on the streets of various cities. Others are operating in closed environments such as mines and industrial sites. Airports have used ‘autonomous trains’ for some time, safely moving thousands of passengers around from one terminal to another. Road testing of city-to-city AV fleets are not far behind. Such vehicles are another in a series of ‘disruptive technologies’ whose benefits are said to outweigh the possible costs. It’s claimed that the current system of independent vehicles driven by fallible humans is so expensive, dangerous and out-dated that it needs to be replaced. At first glance, it’s not hard to see why. Such a system could be more efficient, less wasteful and safer. The outlook appears sufficiently compelling that the longer-term goal of creating fully automated systems is being widely debated and planned for.
Several levels of autonomy are envisaged. At level one single functions are to be carried out by the vehicle in restricted circumstances. At level two the vehicle can operate multiple functions with the driver actively monitoring. At level three the vehicle can cover all driving functions but refer back to the driver if / when needed. At this level, however, the ‘hands off’ issue becomes a safety concern. So at level four, complete vehicle autonomy within system-wide limits becomes the preferred goal. At level five, in-vehicle systems replace all driving functions in any circumstances, indicating true autonomy (King, 2017). Such apparently positive conclusions appear to be supported by World Health Organisation (WHO) statistics that recorded a staggering 1.25 million road deaths in 2015 alone (WHO, 2015). Or, as one writer put it: ‘the only difference between a human driver and a machine driver is the speed and accuracy of perception and reaction, and the machine wins that one easily’ (Walsh, 2016). This is one of several arguments. Others include the following. If AVs were to become standard then chaotic and crowded road transport systems might well be rationalised. Traffic jams could become a thing of the past. Car ownership per se would decline since fewer vehicles would be needed. Roads could be smaller and less intrusive. The space in cities presently devoted to parking would be reduced making these same areas available for other uses. Then again, since the new AVs would run on electric power there’d be an increase in energy efficiency with corresponding reductions in exhaust fumes and pollution. (That noted, the makeup of energy systems – coal, oil, gas, nuclear vs. renewables – used to power electric vehicles would obviously have a significant impact on the overall energy profile.) From a popular viewpoint, cities could return to being ‘clean and green.’ On the other hand all these assumed benefits turn out to be highly contestable. For example, it’s doubtful if such a multi-dimensional transition could occur as quickly as proponents suggest. Then there’s the huge question of costs – not only to manufacture smaller, lighter batteries but also to drive down the cost the sophisticated electronics such vehicles require. Equally, the question of complexity has barely figured in current narratives. But it will take heroic levels of reliability to keep such vehicles operating safely. There’s also another side to this story.
Unemployment and the myth of perpetual Internet reliability
The most obvious and immediate drawback is the rapid decline in employment for large numbers of people who currently earn a living through driving. In the UK, for example, there are close to 300,000 Heavy Goods Vehicle (HGV) drivers alone, most of whose jobs would disappear (Ashley, 2017). And this is without counting bus and taxi drivers. Yet little is heard from policy makers or AV promoters about these deteriorating prospects. John Harris describes the issue like this:
There are 3.5 million truck drivers in the US, as well as 233,000 cab drivers (an official estimate, which seems low), 330,000 Uber drivers and 660,000 bus drivers. In the UK, at the last count, there were 297,600 taxi or private-hire-vehicle driver licenses in England alone, and 600,000 people are registered drivers of heavy goods vehicles. The traditional logic of the job market has made sitting behind a wheel a fallback option – if all else fails, you can always drive a cab. But no more… (Harris, 2016a).
The beginnings of a solution are likely to involve income redistribution on a wide scale. Proposals for a social innovation – a universal basic income (UBI) - to reduce the strain on what Paul Mason calls ‘the precariat’ crop up occasionally but are a long way from being implemented (Mason, 2016). The political will is minimal, the economics challenging and the issues complex. Yet it’s fair to say that little could be further from the minds of those who favour the introduction of AVs. While most are caught up in the ever more unequal distribution of wealth, measures to moderate such extremes are few and far between. These are matters of real public concern. Yet industry innovators, and those who speak for them, remain preoccupied with technical issues. So they don’t view the structural decline in employment and a corresponding rise in public unrest as any concern of theirs. They are focused on capturing as large a slice as possible of emerging markets. So questions like ‘should we do this?’ give way to ‘can we do this, how fast and where?’ Framing issues in such ways certainly simplifies things.
Yet pursuing the single-minded pursuit of ‘innovation’ on the one hand, while ignoring wider consequences on the other, de-legitimises any pretence to objectivity or detachment. Acknowledging and understanding these links therefore becomes a vital public concern. More people would then appreciate the extent to which corporate and social interests have been poorly aligned for many years (Higgs, 2014; Klein, 2007, Klein, 2014; Bakan, 2004). It was suggested above that privileging technological innovation above all else looks increasingly like a dangerous mistake. On the other hand, costs and disruptions can be moderated or prevented if they are detected and publicised in good time. This is obviously one of the key functions of high quality foresight work in the public interest. If and when the political will is found, more equable solutions can emerge.
There is, however, no ready-made solution to what may be the Achilles Heel of all AV systems – their dependence on perpetual Internet integrity. At the very time when key players are preparing for ubiquitous cyber warfare, the faultless continuity of IT-related systems remains a convenient myth. In this view, complexity becomes a social trap and reliable security a delusion. Yet, as things stand, the pragmatic worldview and raw instrumental power of the main players suggests that they will push ahead regardless. They’re uninterested in permission, regulation or negotiating any diversion from the humanly tragic and debased futures they are creating (Harari, 2015).
Systems rationality, artificial intelligence, privacy
Since most governments lack even the rudimentary means to evaluate the emerging tides of new technology – let alone make informed decisions about their social implications – the question of who will take responsibility for large-scale breakdowns, power-outages and disruptions, whether caused by actual accidents or by malign cyber-attacks, remains open. What is clear is that to the extent that AV systems are progressively installed the torrent of data that they’ll require and generate will become too vast and complex for humans to manage. New levels of automation capable of processing vast amounts of ‘big data’ in real time will be needed. Human control over these systems will therefore diminish. Humanity will have taken another step toward the era of ‘systems rationality’ where notions like ‘autonomy’ and ‘choice’ become meaningless.
One option that can be explored as an alternative to a full on ‘big data’ scenario would reflect the difference between artificial intelligence (AI) and Intelligence Amplification (IA). In the former case the goal is to replace human intelligence with machine equivalents, whereas in the latter it is to augment human capabilities. Driver assisted vehicles are not merely less threatening and problematic, they already exist in significant numbers. So it may be possible to explore a similar process of augmenting human capability and, in so doing, bypass some of the hurdles mentioned here. Yet this is by no means a foregone conclusion. Within a ‘growth at all costs’ corporate worldview optimal solutions appear less appealing than grand visions in which limits have little or no place.
Currently we’re a long way from figuring out how society as a whole can begin to deal with the unending flow of data. Effective AV systems would necessarily be designed to eliminate as much uncertainty, ambiguity and choice as possible. It would record the full details of each and every trip, making it possible for anyone with access to know exactly where and when people have been. Unlike with today’s smart phones whose ‘tracking services’ can still be switched off, no such option would be available. Some criminal activities (such as car theft) might decline but at the cost of ratcheting up the level of surveillance to unprecedented levels. One observer sees it this way. He writes:
Shrouded in secrecy, swallowed up by complexity and scale, the world is hurtling toward a new transnational electro-dystopia … Localisation doesn’t matter that much. The Chinese Internet model and the American giant server farms are proof of the dangerous fact that digital automation is inherently coupled with the efficiencies of integrate centralisation and control (Keane, 2015, p. 33).
AVs are safer for whom?
The issue of safety is one of the key drivers behind the emergence of AV technology. Yet the conversation thus far has taken place within an affluent ‘first world ghetto.’ It’s here that the finance is available and the greatest rewards are expected. Yet the closer one looks the more the whole process appears to do with notions of greed than of need. So it’s worth asking a different question – where are these promised new levels of safety most needed? The answer is – in the very places where they are least likely to occur. The WHO (2015) statistics on road deaths make this clear. The following sample is for deaths per 100,000 people in 2013.
Table 1: Road deaths
|Country||Deaths per 100,000 people in 2013|
|Central African Republic||32.4|
|Democratic Republic of Congo||33.2|
If, in this already one-sided technical view, part of the ‘value proposition’ is that ‘human life is valuable therefore we should reduce the road toll’ then it’s clear that countries with the greatest need for technical assistance are the least likely to get it. The unfortunate truth is that there’s little or no profit to be made from poor and destitute nations. Hence the argument about ‘making driving safer’ clearly rests on ‘first world’ privilege. It depends on (a) excluding the poorest nations and (b) therefore ramping up even further the already unsustainable gulf that exists between rich and the poor. So far as the corporates are concerned poor people can continue dying in their thousands so long as they gain access to the most profitable markets. Obscured by the growing chorus of approval for AVs in the rich West this sad reality has been widely overlooked. Yet its antecedents are well understood. They were described a decade ago, for example, in Klein’s detailed account of what she called ‘the rise of disaster capitalism’ (Klein, 2007).
This section has argued that the full costs of any thorough going implementation of AV technology bring with it very significant costs. These include:
- mass unemployment and few serious attempts to deal with it;
- the further erosion of privacy;
- an impossible commitment to the myth of perpetual Internet integrity;
- the assimilation of people, societies and cultures into a world dominated by machines and governed by the abstract demands of ‘systems rationality’; and,
- a further increase in the unsustainable gulf between rich and poor.
Rationales in favour of the rapid implementation of AVs are therefore not as persuasive as they may first appear. It follows that the rush to implementation needs to be slowed down and perhaps halted – at least for a while. This view is partly about values including prudence and compassion. It strongly supports the view expressed in the previous section that new technologies should be seen and understood in their wider contexts. They are not merely ‘stuff’; each has human, social, cultural and geopolitical consequences, and positive, negative and ambiguous outcomes. The arguments and justifications put forward in favour of AVs thus far appear to depict issues in the simplest and most positive ways, obscuring alternatives and understating the wider costs.
High-tech companies have become surprisingly casual about embarking not merely on one or two but a whole series of frankly outrageous projects that, at base, serve to re-shape the world in their own image. But there’s sufficient evidence to take a stand against careless innovation with ramifying social consequences. It’s now clear that a high-tech world fashioned by and for the corporate sector becomes progressively less fit for people (Klein, 2014; Higgs, 2014; Harari, 2015). There are many other alternatives awaiting our collective attention (Alexander & Mcleod, 2014; Rees, 2014; Floyd & Slaughter, 2012).