Posibilidades actuales y futuras para la regulación de la discriminación producida por algoritmos
Year: 2020
Type: dissertation
Abstract: The growing computational capacity of automated data processing systems has led to an increase in their use in all kinds of human activities. These systems can process massive amounts of data and provide very accurate results, helping decision-makers in both the public and private sectors to classify human beings and predict their actions. However, over the past few years, several cases have come to light showing that these systems can generate significant risks to fundamental rights and other democratic values and principles. This thesis focuses, above all, on the damage that the use of algorithms can generate for the rights to equality and non-discrimination. Western societies have been built upon structures that have historically discriminated against the members of certain groups. These structures of discrimination still persist nowadays. Since algorithms are created by human beings and analyse real-world information, it is easy for these systems to integrate such discriminatory structures as well. This thesis looks at how the legal framework for equality and non-discrimination can deal with the harms that algorithms can cause the rights to equality and non-discrimination. It also identifies the shortcomings of this legal framework in addressing the specificities of algorithmic discrimination cases. However, the use of automated systems generates a number of problems that go beyond the risks generated to equality and non-discrimination. All these problems are closely linked to each other and must therefore be analysed in a comprehensive manner. For example, the opacity and complexity that characterises algorithms hinders the chances that individuals may have in defending their rights when they are subjected to automated decision-making procedures, thus limiting individual’s due process rights. To date, data protection regulations have been the main legal instrument aimed towards dealing with the risks and damage caused by the automated processing of personal data. In the absence of other legal instruments specifically designed to address the risks and problems arising from recent technological developments, the data protection legal framework has sought to provide a comprehensive response to all problems arising from the increasing use of new data processing technologies. However, the data protection legal framework has major shortcomings in dealing with the different problems and risks generated by the use of algorithmic systems. This work points out said shortcomings, in particular, with regard to the protection of the rights to equality and non-discrimination. Finally, this work highlights the need for greater regulatory intervention in the public and private use of automated data processing and decision-making systems and justifies the need for said intervention not only by pointing out the harms that these systems can cause to fundamental rights and democratic values and principles, but also by referring to the market failures that are generated surrounding the data services sector. The final part of the thesis makes a series of proposals which, if implemented, should help to overcome some of the shortcomings of the current regulatory framework in the prevention, management and solution of the risks and damage caused by algorithms. All in all, the thesis conveys the need for a complete paradigm shift in the regulatory instruments applicable to algorithms in order to tackle the many challenges that these new technological developments are placing on democratic values.
Author: Soriano Arnanz, Alba
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7024000648
