From an a priori point of view, the relationship between artificial intelligence and the participation of workers through their legal representation in companies has little connection. But, if we approach it from the accumulation and massive processing of data "controlled" by algorithms, a different situation arises, since by means of these algorithms, not only the past is managed, but, increasingly, decisions are made [1], as transcendent as they are: profiles for hiring workers, establishment of work rhythms, assignment of objectives, the distribution of tasks and even the evaluation of the performance of tasks. A wide range of measures that make up the working conditions and before which the legal representation of workers must have knowledge, powers and tools to intervene. How can we let the machines, which do not negotiate or agree, supplant the balance of forces in negotiation and interpersonal agreement?

Article 33 of Law 31/1995, of November 8, 1995, establishes that: "the employer must consult the workers in the adoption of decisions relating to the planning and organization of work in the company and the introduction of new technologies, in everything related to the consequences that these could have for the safety and health of the workers, derived from the choice of equipment, the determination and adequacy of working conditions and the impact of environmental factors on the work".

Therefore, the company's decision must be reported to the legal representation of workers and, in the best of cases, debated and agreed upon, for the benefit of suitable working conditions (social peace in the company).

Artificial intelligence, in the field of occupational risk prevention, with its algorithmic data collection and processing, has several aspects to take into account: on the one hand, it facilitates the achievement of the employer's safety obligations, but on the other hand, it also serves for an "exceptional" control of workers by transforming their activity, including their status and attitude into data. Let us not lose sight of the fact that we are dealing with human beings. In Law 31/1995, of November 8, on Occupational Risk Prevention and in the Royal Legislative Decree 2/2015, of October 23, approving the Revised Text of the Workers' Statute Law, the employer is defined as the obligated subject in the safety and health guarantees, with administrative, labor and even criminal liability for breaches in occupational risk prevention; but, in turn and as stated in Article 20.3. of the Workers' Statute: the employer's power of management shall in all cases respect the consideration due to the dignity of the workers.

Four points for reflection on the conjunction of artificial intelligence and working conditions: artificial intelligence in the determination of liability for damages, artificial intelligence in relation to employability, artificial intelligence and the establishment of working conditions versus routine jobs, and finally artificial intelligence and the "intelligent" monitoring of work activity.

In the first point, as measures favoring safety conditions at work, there is the identification of potential risk situations, through the probabilistic analysis of causes that can trigger damage, in order to implement preventive measures. In addition to "to causes", this analysis can also focus on the study of behaviors involving the monitoring of workers and their "infractions", in order to "mitigate" corporate responsibilities. To enhance or mitigate the attribution of corporate liability, which is defined by the standard and also by the "interpretation" made of it. For example, in order to establish liability for the crimes of articles 316, 317 and 318 of Law 10/1995, of November 23, 1995, of the Penal Code, it is necessary to go to the labor law to define the corporate conduct, whether or not it is intentional, and thus derive the classification as a crime. In the investigation of an accident at work or in the cause of damages due to exposure at work, it is possible to try to identify from a breach of the employer's duty of safety, through an unsafe negligent conduct of the worker, to a combination of both or even to consider a "fortuitous event". In this task, data analysis by means of artificial intelligence can tip the balance in one direction or the other, depending on the programming that is done.

In addition to the investigation of the consequences of the exposures, we can find in the field of artificial intelligence aspects of transcendence such as employment itself. It is important that the pronouncements of the Social Courts resolve, in the worst case, as "unfair" the dismissals of people who are replaced by machines in the performance of the work (it would be desirable that it would be "null and void"), when the companies allege the mechanization as "technical, organizational and productive objective causes" that support the dismissal. The case of Social Court No. 10 of Las Palmas de Gran Canaria, which ruled that an objective dismissal by a multinational tourism company that intended to replace a worker with a computer program or management "bot" was unjustified [2]. In this sense, an article in "Forbes" [3] denounces that a large number of dismissals in large technological companies are due to the fact that the activity of people is being replaced by artificial intelligence systems. Therefore, a major cause of concern for the legal representation of workers in companies is the permanence in employment and the danger of objective dismissals with a technological cause to cover up the replacement of people by tools derived from artificial intelligence. The replacement of a person or a group of persons by a robot or software has gone from being anecdotal to a socio-labor problem in the context of the digitalization of work. If necessary, a "sustainable technological transition" requires action on the reform of the causes of objective dismissal for technological measures and the establishment of a legal environment of "reinforced" protection for people, with actions on training in those groups and levels most sensitive to the emergence of new technologies in production processes.

A third aspect to consider is how to take advantage of artificial intelligence to improve working conditions. Artificial intelligence should take over routine and monotonous administrative tasks, as long as the person is not replaced, but offered the alternative of more challenging jobs, of greater interest and requiring more professional and "emotional" skills. Training programs that take into account specific needs must be customized and offered. Artificial intelligence can help identify and address problems and solutions in real time, offering "intelligent virtual assistants". But, for these tasks, artificial intelligence has to feed on sources such as e-mails, chat logs and social networks, to assess feelings and identify areas of concern; beware, these are areas that invade the privacy of working people. Therefore, the balance of the intervention must be agreed upon.

The fourth point is to see who organizes the work and on what criteria it is based [4]. It is not uncommon for "smart watches" to be proposed for working people. They are portable, rugged devices that can withstand wear and tear while working. They include health monitoring functions; that is, they serve to monitor health metrics such as heart rate, blood oxygen level monitoring, blood pressure monitoring, alerts on abnormally decreased or increased heart rate; as well as a stress monitoring function to detect changes in indicators of physical and psychological well-being; noise pollution monitoring; also sleep studies and interrupted sleep patterns; in turn, the watches can include "life-saving" detection systems for isolated work or work at height, so that after a strong impact, if there is no interaction with the watch, it automatically contacts an emergency point to which it transfers the location; they also include GPS tracking. Among other functions the smart watch alerts when a call is received, a text message or a calendar notification; in addition the smart watch can contain weather alerts, synchronized in the environmental agency that is programmed, emergency button; also alarms for reminders, alerts... In the same device are security elementsthatcan also be used as elements for the control and daily monitoring of work and personal activity and their individual fitness. Where is the limit. Again, the barrier between security and intrusiveness lies in who sets the parameters for artificial intelligence to act and what parameters are considered "optimal" in work performance. Artificial intelligence does not work without programming, and workers must actively participate in this programming through their legal representatives.

In conclusion, the undoubted advantages of the use of automated decision-making systems based on algorithms cannot ignore the risks that they entail for the rights and interests of workers; the main one is that the person becomes a mere appendix object of the computer programs. To avoid this, it is essential that there is effective participation and control by workers and their union representatives to ensure that there is an adequate supply of information on the algorithms used. Moreover, all algorithms used in companies that affect working conditions should be supervised by the "labor and digital authority", specifically in labor matters, to see if they take into account the regulations and the rights enshrined in a fundamental right, such as health and safety. In other words, it must be ensured that there is a right to know the algorithms and a supervision of the same, of such a nature that it can be considered integral to the content materially protected by the fundamental right to the protection of personal data.

There has already been a pronouncement regarding the requirements of the workers on the elaboration of the risk map, after the evaluation of jobs, where the methodology, procedures and algorithms that are currently used in reference to the working conditions of the staff, working hours and workloads are established and identified, as these issues represent an element that represents a negative and significant influence in the generation and analysis of risks [5].

Also to the order allocation system in the automatic allocation system that is performed telematically by the GLOVO algorithm, following a cost-benefit function that seeks the best possible order-delivery combination that minimizes the sum of costs [6].

In turn, VODAFONE, SAU and VODAFONE ONO, SAU have been reducing the services required to INDRA based on the result of the NPS survey, which determines the quality of service provision. The aforementioned companies have an application, which combines, by means of the corresponding algorithm, the distribution of the workload among its subcontractors, through an application, which activates a target device, which matches call forecasts, response to peak demand and call recovery time, so that, if relevant delays occur, it is distributed to other providers [7].

In conclusion:

The participation of workers through their legal representation is not only a fundamental element for the design of working conditions and guarantees for the health and integrity of workers, but also has the regulatory support for it. It is necessary to establish procedures to audit the health results of the application of the algorithm or the application of Artificial Intelligence. If it is a question of defining an algorithm, the variables that constitute it, as well as the measurement and evaluation criteria, must be shared between the management and the workers' representatives; avoiding that it is offered by the employer as an unquestionable mechanized mathematical process, an example of good practices in this sense is the one offered in JUST EAT [8]. The detailing of the algorithm must have an impact on aspects such as non-discrimination, transparency, responsibility, proportionality, impartiality, access to data and, above all, ensuring that it is people who continue to control and be accountable for decisions in the workplace. For the trade union side to maintain the correlation of forces. The idea is that, while new technologies offer both potential advantages and disadvantages to workers, trade unions can and should negotiate agreements on the introduction, implementation and governance of any algorithmic management tools in the workplace. Neither teleworking nor algorithms can be a setback in labor rights and even less in the prevention of occupational hazards, technological advances have to improve the lives of everyone, employers and workers. Technological changes should not be denied, we are not going to burn the factory, the task of trade unionism is to govern it, to introduce mechanisms so that it does not mean a setback and to adopt agreements to ensure its control. In addition, we must have the public guarantee of a supervision, by the competent entities, to strengthen the rights of the weakest part of the labor relationship.

[1] Finn, E.; <<La búsqueda del algoritmo: imaginación en la era de la informática>>, Ediciones Alpha Decay, Barcelona, 2018.

[2] Judgment of the Social Court of Las Palmas, on September 23, 2019 - Speaker: JAVIER ERCILLA GARCIA. No. Appeal 470/2019. Termination of the contract of an employee of a multinational tourism company of Gran Canaria origin who was dismissed, after 13 years working as an administrative employee, to be replaced by a computer program, or management bot.

[3] Artículo: <<¿Por qué los despidos de Google deberían preocuparnos a todos? Más de 95.000 trabajadores del sector tecnológico fueron despedidos en 2023. Casi a diario las principales empresas, como Meta, Amazon, Salesforce y Micrososft, anuncian recortes.

[4] Todolí Signes, A.; <<Algoritmos productivos y extractivos. Cómo regular la digitalización para mejorar el empleo e incentivar la innovación>>. Aranzadi, 2023.

[5] Supreme Court Ruling 363/2022.. Social Chamber. Resolution nº 50/2022. Speaker: ANGEL ANTONIO BLASCO PELLICER. Appeal No.: 205/2021.

[6] Supreme Court Ruling 2924/2020.. Social Chamber. Resolution nº 805/2020. Speaker: JUAN MOLINS GARCIA-ATANCE. Appeal No.: 4746/2019.

[7] Supreme Court Judgment 3166/2017.. Social Chamber. Resolution nº 624/2017. Speaker: ANTONIO VICENTE SEMPERE NAVARRO. Appeal No.: 25/2017.

[8] See minutes of agreement SIMA - FSP Just Eat File no. M/441/2021/N.