Cyberboss: Here’s how AI is reorganising the lines of class struggle

There is a growing sense that the future of work might not unfold in our favour. People are expected to work longer, for less, with less security and fewer protections. Rather than making work easier or more rewarding, we expect the development and application of new technologies, particularly in the areas of automation, computation and artificial intelligence, to disempower us.

Concerns around the degradation of work are not new, but our everyday experience of moving through the world over the last ten or fifteen years tells us that many of the jobs that are now being created are less secure, more stringently managed and paid worse relative to the cost of living than ever before.

It is commonly understood that this world of work has some relation to the proliferation of computational or algorithmic technologies, which are being applied to work in novel ways.

Indeed, digital technologies appear to be changing the world of work at a fundamental level. Left unchecked, they may well lead to forms of work that are increasingly stressful, injurious and dehumanising.

The archetypal example of a modern workplace where these concerns are at stake is the Amazon warehouse – or ‘fulfilment centres’, as the company calls them. In these facilities, poor employment protections, high workloads and advanced technology conspire to create high-turnover jobs that come with a harsh toll of mental and physical exhaustion. But such conditions are not confined to Amazon.

They are widely acknowledged features of the so-called ‘gig economy’ as a whole, from transportation platforms like Uber and Lyft to food-delivery platforms such as Deliveroo in Britain or GrubHub in the United States. What unites the Deliveroo rider’s app with the largest Amazon fulfilment centre is a shared technology.

I call it algorithmic management. It is a simple idea with big implications, made possible by the vast computational power now available to rich companies. Algorithmic management refers to a way of organising work in which workers are directed, monitored, tracked and assessed – all at once, in real time – by a computer system that doesn’t rely on a human manager to control it. Far beyond its application in headline-grabbing companies such as Amazon, it is a way of organising work with wide-ranging consequences.

But aren’t robots coming for our jobs? Beyond the PR of the wealthiest companies that can afford the most self-aggrandising research and develop- ment projects, this is generally not the case. I wouldn’t be so foolish as to predict that unmanned delivery robots, Amazon drones and workerless ‘lights-out’ factories will never materialise, but for the moment the application of automation to work generally takes place in discrete sections of the work process – typically at the level of decision-making, rather than through the replacement of workers altogether.

Aside from being costly to invent and acquire, robots need to be maintained, cleaned and fixed. Experiments with mobile robots in distribution warehouses quickly ran into the rather mundane problem of dust. While ‘workerless’ factories have been predicted since at least the 1970s, I would gently suggest we’ll only really begin to see them in large numbers about a decade after companies master the ‘dustless’ warehouse. It’s not impossible, but it’s improbable.

Even in the oft-cited example of the supermarket self-service checkout, it’s not that the work of scanning has actually been automated – it has simply been transferred from a paid employee to the paying customer. The fact is that, for most practical work, human workers are simply cheaper, more reliable and easier to replace than robots. Rather than being replaced by computers, it is instead the case that ever more workers are being managed by them, by virtue of workers being subject to algorithms – computational codes, or rules, that once executed convert data into decisions.

Rather than being replaced by computers, it is instead the case that workers are being managed by them

Just as a calculator (just one type of computing machine) can outstrip mental arithmetic, algorithms can do things that people simply cannot. They can draw on an extraordinary range of information, decide and act in real time, monitor and measure with a high degree of accuracy, close ambiguities, work constantly, interact with users dispassionately, and – if necessary – archive, or ‘remember’, everything.

Within a work context, they can integrate a wide range of modular technologies (especially for tracking), connect physical actions to digital records, mediate between different types of information, and make calculations of efficiency based on a range of variables.

For a human this work is complex and time-consuming, if it is possible at all; for a computer system it is both quick and easily repeatable. Moreover, and not insignificantly, calculations produced by computers attract an air of infallibility or objectivity among human users. In the ideal algorithmic management form, replaceable workers work on the basis of algorithmic decisions, and are held to ‘objective’ standards set by algorithmic analytics. This is the real future of automation. It is about workers and algorithms.

But it is really about power; and like most meaningful reflections on algorithms, it is really about communication and control – which is to say, cybernetics. Cybernetics is an old term that in modern usage is best thought of as the science, study or conceptualisation of systems of communication and control.

The most widely understood principle given to us by cybernetics is the idea of the feedback loop: a circuit or process in which some output is ‘fed back’ into a system as input, typically with the aim of correcting or optimising that system.

For our purposes, what cybernetics is interested in is the ability to use feedback loops to control or steer a complex system (in this case, a workplace). You do not need a technical knowledge of computers, software or code to understand these concepts. It will be enough merely to have some idea that computational technology now shapes our everyday experience of the world in extensive ways, from smartphones and the screens that are becoming ubiquitous in public spaces to Bluetooth and barcodes.

Similarly, you do not need to have ever worked with an algorithm, though the chances are you already have, knowingly or not. It is enough that you have an interest in workers and their working lives, and care what the future holds for them – for us. So workers are being managed by computers rather than replaced by them, and those computers are organised on the basis of cybernetic feedback loops that control or steer the workplace. Can workers fight back? The answer to that is not straightforward.

Trade unions

In the Anglophone world, trade unions have been the primary vehicle of working-class power for at least the last hundred years. There are trade unions for almost every line of work we can imagine, and an increasing number of general unions that cast a wide net over lots of different types of workplace, particularly those in the private sector. Within the political and economic system in which we live – capitalism – trade unions are basically a good thing.

They mediate between labour (workers) and the forces of capital (employers, owners and, where necessary, government), and they are largely responsible for working norms like paid annual leave, the five-day week, the eight-hour day, child labour laws, and health and safety legislation.

Yet trade unions are not the same as working-class power, especially in the context of algorithmic management. In the contemporary economy, there is no doubt that unions continue to be valuable, and it is true that, in both Britain and the United States, the trade union movement has been beset by the repressive and hateful anti-union agendas of successive governments since at least the 1980s.

I make no concession to such laws, which are designed to stifle the ability of working-class people to express themselves politically within the world of work. But there should be no doubt that there are better and worse unions in terms of ambition, strategy and political will, and that within even the most impressive unions there are both proactive and reactionary factions, as anyone who has ever participated in union organising can attest.

But where technology is concerned, trade unions remain on the back foot, even without the assistance of the political opponents of the working class. Or, perhaps more generously, the distance trade unions have to cover to meet the demands of the present moment depends on their historical relationship to questions concerning technology and the work process. Yet the continued application of digital technology to work, and workers more generally, is a future we are all going to be forced to confront one way or another.

In a world where the vast majority of us are compelled to work in order to live, the future of work matters – which is to say: workers matter. It matters whether people spend half their waking lives happy or stressed; it matters whether we feel ourselves to be dehumanised or dignified; it matters whether workers are powerless, or – as the case may be – not.

In a world where the vast majority of us are compelled to work in order to live, the future of work matters: workers matter

This means the conditions that organise our work should always be a political priority, at least as long as life and society are organised around work in turn.

Algorithms are now the lifeblood of industrial innovation, and algorithmic management represents a precipice from which we will struggle to return if we do not take it seriously.

Unfortunately, so-called ‘human-centred’ approaches to this problem offer little in this regard. Humans – and, more significantly, politics – are already at the centre of the algorithm. Nor is there any amount of job security or mending wages that will mitigate algorithmic management itself.

What we need, therefore, is a political understanding of algorithmic management on its own terms. The task of interrogating workplace technologies should be fixated less on what can be done to make them ‘fairer’ within parameters acceptable to the parties implementing them – a paternalistic, conservative posture that is too often adopted uncritically – and more engaged with the questions of power that led to injurious technologies being introduced into workplaces (itself a euphemism – we ought to say imposed upon workers) in the first place. This means starting with algorithmic management as we find it.

By necessity, where there are workers, there is also politics. Workers are variously unpredictable, fallible, collaborative, creative, lazy, cooperative, uncooperative, and much else besides.

In a word, they are free, and by their most basic and human capacity for autonomy introduce uncertainty into a finely calibrated calculus for making money. If one of the applications of digital technology to work and workers is the reduction of this uncertainty, it is a fundamentally political set of concerns – ones not merely reducible to their economic rationale – that is at stake in algorithmic management and the digitalisation of the workplace.

*This is an extract from the new book Cyberboss by Craig Gent

openDemocracy has teamed up with bookshop.org, an online retailer that supports small independent bookstores. When you make a purchase through a link on our page we will receive a small commission on the sale.