The use of algorithms in labor relations enters the agenda of unions

A few weeks ago, the research and studies section of the Spanish union UGT published a document under the title Algorithmic Decisions in Labor Relations in which they propose two lines of action in front of the generalization of the use of algorithms in the field of people management in companies.

On the one hand, the promulgation of a ‘Law of algorithmic justice in labor relations’ at both Spanish and European level that regulates the use of this type of technological solutions in decision-making regarding selection of new employees, promotions, work organization, performance evaluation, calculation of remuneration or application of disciplinary measures, among other dimensions of people management.

On the other hand, they propose that this subject of algorithmic management be subject to collective bargaining and set some premises and guidelines for future negotiations on this matter.

Towards a law of algorithmic justice in labor relations?

The document enumerates the premises of a future Law of Algorithmic Justice in Labor Relations. Among them, I would highlight the following two:

First, extend the application of article 22.1 of the General Data Protection Regulation to any aspect of the employment relationship. This article establishes that “the data subject shall have the right not to be subject to a decision based solely on automated processing, including profiling, which produces legal effects concerning him or her or similarly significantly affects him or her.” Specifically, the union envisions a law that a) enshrines the right of workers to choose not to be the object of algorithmic decisions, and b) impedes workers are discriminated against or have their socio-labor conditions reduced for exercising such rights. To do so, UGT proposes to set a Registry of Labor Algorithms in all companies. All algorithmic solutions “that affect the organization of work” would be listed in that Register that could be consulted by both workers and their representatives.

Second, they demand all algorithmic decisions that have an impact on workers to be audited. Those audits would accredit and certify that algorithmic decisions are independent of any bias in the input data set and verify that the final results are objective, truthful, fair, and accurate. They would also verify that those decisions respect the rights and freedoms of workers. In this regard, UGT proposes the Labor Inspection be responsible for these audits. For this, the Government will provide the Inspection “with the necessary resources to be able to carry out these audits, creating specialized units for this purpose, which will have the legal authority to verify the suitability of the performance of any algorithm applicable in labor relations.”

Algorithmic decisions in collective bargaining

Regarding collective bargaining, although the labor implications of new technologies are not a frequent topic in negotiations (in 2020, only 6% of collective agreements in Spain included clauses on “implementation of new technologies”), UGT aims at including the algorithmic management of workers within the topics subject to collective bargaining.

To guide negotiations in this area, the union proposes in its document a series of premises and guidelines inspired, on the one hand, by the resolution of the European Trade Union Confederation on artificial intelligence and data and, on the other, by the Algorithmic Management Guide produced by UNI Global Union to guide negotiations on the development and application of algorithmic management tools in workplaces around the world.

Among the premises for negotiations UGT includes, for example, the need to make employers reflect on the real reasons that lead them to introduce algorithmic management in labor relations, given the risk this decision is the consequence of a fad and not so much a thoughtful decision, with clear objectives and an explainable logic.

The document also raises the principle that, since employers are always responsible for the decisions that affect their workers, and must always obey the law, employers, in no case, may use the use of algorithms as an excuse to circumvent the responsibility for their actions. That is, the argument of “is what the algorithm says” should not be acceptable. Algorithms can provide advice, but it’s up to people to decide, and “one human at the end of the chain of command” should be a guiding principle in all cases. However, this “human in command” principle must be authentic. The person at the end of the decision chain cannot just endorse what the algorithm says.

Finally, I also find it remarkable what the document states on the ‘surplus-value of data’ that can derive from using algorithms to manage people. The union argues that data (and the algorithms that use that data) can generate a real economic and competitive benefit for companies, from which workers should also benefit. In their opinion, we must keep in mind that data, in their genesis, belong to each worker. Therefore, the logical thing is that the “dividend” generated by the use of algorithms to manage employees also benefits the workers who produce or provide that “raw material” (their data) essential to this new source of business benefits.

Time will tell us if the law on algorithmic justice in labor relations becomes a reality and if algorithmic decisions in people management end up being a common subject in collective labor agreements. In Spain, as we pointed out before, we start from far behind. The percentage of collective labor agreements with clauses on the “introduction of new technologies” was barely 6% in 2020 and 5% in 2019. But it is precisely for this reason that we should welcome a document that brings unions closer to an emerging challenge of the future of work that they had left a bit unattended.