The Spanish Ministry of Labor interprets the scope of company obligations to provide information on algorithms
Spain Labor and Employment Alert
In guidelines published by the Ministry, it details the information that must be provided by the company to the workers’ representatives and to the workers themselves. The document includes a tool/questionnaire to determine and systemize the obligations relating to the use of algorithms.
The Ministry of Labor has published guidelines, which are not regulations, called Algorithmic data in a labor environment, that cover the obligations and rights relating to the algorithmic data that companies are required to provide to the worker's legal representatives and the workers themselves, pursuant to article 64.4.d) of the Workers Statute (WS) and article 22 of the Data Protection General Regulations (RGPD). The guidelines also contain a tool to determine and systemize these obligations.
According to the guidelines, the information obligations have a dual dimension:
- Individual rights. Article 22 of the RGPD includes an obligation to provide information to the workers subject to fully automated decisions without human intervention, including the preparation of profiles.
To understand that human intervention exists, it must be significant, carried out by a person with the authority to make decisions and assess all the available information. Human intervention must be observed in each one of the decisions and not only in part of the process. When human intervention is limited to repeating the decision reached by the algorithm, it cannot be understood as significant intervention and therefore as a fully automated decision-making process. - Collective right to information. Since the entry into force of the so-called Rider Law, article 64.4.d) of the WS has incorporated obligations to provide information to the worker's legal representatives on the use of algorithms and artificial intelligence systems affecting the decision-making process that may have an impact on working conditions, access to and maintenance of employment, including the preparation of profiles. In this case, the decision does not have to be entirely automated for the information duty to exist.
The guidelines interpret that both obligations have a minimum content, referring to the information to be provided by the company, which consists of:
1. Information on the use of algorithms and artificial intelligence systems for automated decisions, including the preparation of profiles, identification of the technology used and the management decisions reached by persons in relation to the use of such technology, which include:
(i) The use of algorithms and artificial intelligence systems to reach management decisions on personnel and prepare personal profiles for application in a labor environment, (ii) the specific personnel management decisions that are reached using algorithms or automated decision-making systems (selection of personnel, hiring, assigning of tasks, control of productivity, promotion, etc.), (iii) the type of technology used by the algorithm if it generates a "black box" algorithm (when the procedures, logic and variables are not immediately evident or known) or if it is an ongoing learning algorithm, (iv) the specific software or product used and, if applicable, whether it has any type of certification, and (v) the degree of qualified human intervention in the decisions reached.
2. Significant, clear and simple information on the logic and functioning of the algorithm, including:
(i) The variables and parameters used, such as the type of profiles prepared by the algorithm, the variables used, understood as the information or factors used, (ii) the parameters used by the algorithm to reach automated decisions, understood as the relative weighting of each variable, (iii) the rules and instructions used by the algorithm, understood as the programming rules, (iv) the training data, (v) the accuracy or error matrix in the automated tasks, and (vi) the audits carried out or assessment of the impact by the company.
3. Information on the consequences that may result from the decision reached using algorithms or automated decision-making systems, such as:
(i) The consequences of the decision reached using algorithms or automated decision-making systems on the person and (ii) in addition, the worker's legal representatives must be provided with information relating to the effects on equality and non-discrimination between men and women.
Other relevant issues relating to the interpretation by the guidelines of the algorithmic information duties are:
- The information must be provided in a clear, simple and understandable way for people without technical knowledge and the provision of merely technical information is not sufficient and may therefore be opaque or confusing. The obligation to provide information cannot be interpreted as the provision of the source code of the algorithm.
- Given the lack of regulatory accuracy relating to the concept of algorithms, any algorithm used in a labor environment to reach automated decisions, including the preparation of profiles, is subject to the corporate obligation to provide information.
- This information obligation includes all the decisions relating to personnel management, from selection and hiring to the assigning of tasks, establishing of timetables, determining of salaries, control and monitoring, control of productivity, assessment of results, professional promotion and even dismissal.
- The moment at which the information must be provided should be:
- To all workers individually affected by fully automated decisions, prior to the processing of data; in other words, prior to the preparation of a profile or automated decision.
- To the workers’ representatives with the correct periodicity, which is interpreted as prior to the use of an algorithm and any change in the variables, parameters or any other feature thereof.
- As a general rule, there is no obligation to negotiate the algorithm with the worker's legal representatives. Nevertheless, there does exist an obligation to negotiate when the algorithm is used for collective dismissal, for example, to determine the persons affected by the dismissal or other collective measures.
- Despite the fact that article 64.4 WS only establishes an information obligation, the guidelines interpret that there is a corporate obligation to consult the use of algorithms for personnel management, pursuant to article 64.5 WS (which refers to the right of the worker's legal representatives to be informed and consulted on all company decisions that may give rise to relevant changes in the organization of work and employment contracts). In addition, it recalls the right of the worker representatives to issue a report prior to the "implementation and review of systems of organization and control of work, studies on times, establishing of premiums and incentive systems and the assessment of positions" by the company.
The guidelines also include a tool/questionnaire aimed at specifying and systemizing the information obligations relating to the use of algorithms and automated decision-making systems in a labor environment.
Finally, upon presentation of the guidelines, the imminent implementation of the algorithm to pursue non-remunerated overtime was announced. The MAX algorithm ("more algorithms for less overtime") is aimed at limiting excess working hours and, above all, uncontrolled and non-remunerated overtime.
Contact