Artificial intelligence and algorithms are increasingly involved in the work environment. They already read resumes or assess productivity. The benefits they offer to workers when looking for work are also sold. On the other hand, will they also decide on layoffs? In Europe it has marked some exit and its implications in Spain raise doubts.
AI advances. In companies with more than 250 workers, almost a third use it to direct workers, explains Adrián Todolí, professor of Labor Law at the University of Valencia. Provides support when scheduling, hiring and selecting workers. Decides work rhythms, makes performance rankings. For example, among maids in hotels: if they take too long, they are singled out and it serves as an argument for dismissal. It also marks the routes of the riders and penalizes those who are less willing or less qualified.
To be fair, the algorithm and AI must meet three critical elements, says Pedro César Martínez, director of the Human Resources master's degree at Comillas Icade. “That they are transparent and explainable, that they respect privacy and that they avoid discrimination and bias.” The first is tricky. The algorithm will indicate that that person should be fired, but it will not go into the reasons. It will only be based on measurements and piles of data. “The worker has to know where he is and what is asked of him,” continues the expert.
The main doubt is whether a correct justification is achieved and therefore the subsequent validity of the dismissal. “The worker could appeal due to the lack of understanding of what has happened, he understands that it is unfair and that there is a lack of transparency and will go to court,” explains Todolí. An artificial intelligence analyzes data and knows how to pair the best workers in a shift, but it does not know why they are a good tandem. In the same way he doesn't know why they are worse or have to leave. And problems can arise due to poor configuration, such as the drop in productivity not taking into account that a worker is sick, on leave due to the death of a close family member, or having unforeseen events while doing the work. “If your boss is human this doesn't happen. You can't talk to the algorithm, it is more difficult to solve an error. It dehumanizes work and reduces empathy,” says Todolí. “By not being able to justify the decisions, many times you will not be able to complain,” he warns.
Having someone in the flesh make the final decision and make sense of the data would be the solution, Martínez believes. Without blind trust. “We cannot support dismissal due to AI, it must be another element when making decisions. A dismissal is not arbitrary, it must be justified and supported, whoever makes the decision must argue it correctly,” she completes.
Lack of clarity can lead to more disputes. As an example, Todolí cites the case of Poland, where several rulings have declared dismissals determined by a machine to be illegal. They argue that the algorithm changes the objectives every day or week, something that does not fit with a fixed contract, which is always the same, with clear and non-variable objectives. Also in that there is a lack of transparency. "We don't know how it calculates productivity, objectives...". “It is perfectly applicable to Spain, the arguments are transversal legal principles,” says Todolí, who has just launched the book Productive and Extractive Algorithms (Ed. Aranzadi).
The judicial decision may bring more queue. If the dismissal is unfair, the company could face compensation and continue operating with the algorithm. The solution is for the courts to declare them null and void, which forces the employee to be reinstated, or for the compensation to be higher to discourage misuse. Or have a step prior to justice of response and explanation.
To put things in order, we are awaiting the regulations that are being finalized at community level. The danger is that it is born already outdated.
