Using AI to Help Fight for Workers’ Rights

Victor Figueroa
Sep 6, 2018 · 7 min read

For the last 40 years trade unions across much of the world have been on the back foot. Neoliberal policies have shrunk the state, reduced investment in infrastructure and hammered labour rights. The result is a growth in inequality, and increasing levels of informal employment. Union membership is generally declining, and the age of union members is increasing. In some countries, particularly those in what we might call the ‘anglosphere’, the English-speaking world, the picture looks grim for workers.

Unions are trying to develop new ways of organising, and new forms of campaigning in an attempt to stem this decline, with some success it has to be said. But so far this success has been limited by an overarching context that acts against unions.

One response is to take a more political stance than before — for unions to become more like social movements — less concerned with the narrow issues of pay and conditions, and more engaged with the broader economic and political issues that affect workers. In fact, in the face of rising far right mobilisation in the developed world this is essential. Some unions — like ITF affiliates the RMT and Unite the Union in the UK — are working in this direction — and it is bearing fruit, just as it did in the past.

But there is another way that unions could enhance the effect of both their organising and their political campaigning, and it involves technology.

Modelling disputes

The more we know about a situation the more likely we are to resolve it successfully. Traditionally, in a dispute with management, unions try to develop a good understanding of a company, of its management and their concerns and priorities. They also make use of their understanding of the workers and their issues, and sometimes the situation is placed in a broader context. This information helps inform decision-making. At some level, before a decision is made about how to confront an issue, the various options are discussed and categorised. Eventually a decision is made, and the workers take action, such as a strike.

The basic preconditions for a successful action are well known by both trade unions and management — high membership levels, unity, workers having a clear goal and understanding of the reason for the dispute and potential costs, good leaders, knowing the impact of the action on the company and more broadly, good communications and external networks and so on.

But the decision-making process around a dispute is as yet still relatively unsystematised. Yes, there are theories, and there is often a lot of experience in the room, but at the same time, when looking to assess the potential success of the action, it is really experience, common-sense and gut instinct that guide the decision.

At root industrial action by workers is a form of confrontation between workers and management. In some parts of the world this confrontation is softened, or made unecessary by tripartite mechanisms that ensure workers have a say. In these fortunate places workers can discuss their issues with management without need for confrontation. Dialogue prevents conflict.

But in many parts of the world these mechanisms don’t exist, or they are very weak. And since management can usually count on a friendly legal system, as well as upon the services of the police and other state institutions to repress workers, they have little incentive for dialogue. This means that workers’ are forced to take up disputes with management as a confrontation. Workers usually have no other way of trying to get management to listen to them.

Looked at in this way, an industrial dispute is a form of conflict. Like a war, which is what happens when a dispute between states results in violent conflict.

Why am I drawing this parallel? Because the military world also deals with conflict. Unions have recognised this in their adoption of the language of war in relation to ‘campaigns’, ‘tactics’ and ‘strategy’. Militaries around the world have gamed conflicts for a long time, in order to prepare and train people, and assess possibilities.

Exercises and war games, and increasingly, simulations, are common in preparing the military for the ‘real thing’. These activities, within the framework of a strategy and a set of tactics, don’t replace decision-makers, they help them assess whether a battle is worth fighting, and help guide them in how to fight it.

Is it not logical for unions to adopt this practice?

It is difficult, if not impossible, for unions to get workers to carry out training exercises in the way the military can. Unions do not control workers’ time. Disputes are governed by restrictive laws, and have a real impact on peoples’ lives and the economy. Moreover, union resources are usually scarce.

However, new technologies, particularly developments in Artificial Intelligence and the reduction in the costs of computing power make it increasingly clear that the simulation, or modelling, of industrial disputes and legal cases should start being used by trade unions.

These simulations can be used both to train shop stewards and organisers, and to model the outcomes of disputes.

Imagine it, a dispute arises in a company. Union officials turn to their simulator and run different scenarios after inputting variables. What would happen if we take action in location A, as against location B? Which parts of the workforce would be key? How would the public react to different types of action? All of this could be modelled. The union officials could then use the simulator to compare outcomes, and produce an analysis of the probability of success depending on the sequence of actions taken. This could then help them to choose the best option.

At first the system would produce quite rough results, but as time went on it would build a database of its own, and more and more data about different companies could be uploaded to it. The actual results of campaigns could be inputted, helping the AI to learn from its past mistakes, eventually providing more accurate simulations.

Using AI like this would not replace the experienced organiser, but it would enhance their ability to make the right decisions.

Developments in AI potentially mean that such simulations could ‘learn’ from broader and broader datasets, eventually becoming an electronic archive of knowledge about companies and industrial disputes — and crucially, able to turn that experience into probabilities.

Modelling policy

Eventually, as AI develops it would even be possible to build models of industrial sectors, and national economies, which would develop more and more accurate scenarios, potentially even providing data about where unions have gone wrong in their disputes.

While it may seem far-fetched at the moment, some companies are beginning to use modelling to feed into their decision-making processes, learning from the financial sector. At some point they may begin to use them against trade unions, enhancing their already formidable ability to win disputes.

The global trend is towards the relatively (and increasingly) cheap option of electronic simulation to inform action in the real world. For example, simulations are increasingly used in politics to model electoral outcomes. China is also using AI to help its diplomats make decisions. Companies, governments, militaries and civil society organisations will increasingly use AI and digital models to inform their actions and help achieve their goals. Workers must not be left behind in this process.

If the right data could be collected, such union simulations could be expanded to cover an entire country — its economy, political system, media networks and its social structure. This information would then provide a realistic model on which to work through all aspects of a dispute, and potentially enable unions to also model political campaigns, economic policies and so on.

It does not take much imagination to foresee that such simulations would have important political effects. Any proposal, any policy could be analysed by a model, and its results predicted. If applied by unions, it would substantially enhance public awareness of the real impacts of policy. It would substantially enhance our decision-making by allowing unions to ally tacit knowledge and hard-won experience to data and probabilities.

It won’t replace organising and campaigning. It won’t replace political lobbying. But it will hugely enhance workers’ ability to win their struggles.

If global federations developed models based on regional variations it could be used to help unions in the most anti-union environments of the developing world. It would therefore benefit workers across the world.

But in order to build a capacity for simulation unions need to start putting resources into the necessary hardware and software. We need the right computers, and the right programmes. We need to gather the right data from our members, and other sources, and develop the right mathematical models. The specialists are out there, but at the moment nobody is offering them the incentive to work for trade unions.

It probably wouldn’t take much. Offer developers a couple of million dollars prize money and no doubt teams of engineers will start working on the problem immediately.

Unions will begin using AI eventually, but the earlier it happens the better we will be able to fight for social justice, democracy and workers’ rights.

Victor Figueroa

Written by

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade