Transparencia algorítmica ¿un problema normativo o tecnológico?
Transparencia algorítmica ¿un problema normativo o tecnológico?
Authors
Araya Paz, Carlos
Authors
Date
2021
Datos de publicación:
Cultura, Hombre, Sociedad (CUHSO), Vol.31, N°2, 306-334, 2021
Keywords
Transparencia - Algoritmo - Sesgos - Aprendizaje automático - Regulación
Collections
Abstract
Con la evolución de las tecnologías, cada vez va a ser más frecuen-te que ciertas decisiones tales como selección de personal, otorgamiento de crédito o selección de parejas, sean adoptadas por algoritmos de aprendizaje automático. Muchas de esas decisiones pueden contener sesgos que derivan en decisiones discriminatorias que afectan al sujeto sobre el que recae dicha decisión. Para revelar cómo se adoptaron dichas decisiones, se ha recurrido a la idea de transparencia en materia de algoritmos, implementándola a través de regu-laciones. Dicho enfoque normativo o regulatorio ignora una premisa básica: la transparencia no es solamente un problema legal, sino también un proble-ma tecnológico, que adquiere relevancia cuando hablamos de algoritmos de aprendizaje automático. Los algoritmos de aprendizaje automático constituyen la base para el de-sarrollo de una tecnología idónea para automatizar procesos más sofisticados que el enfoque clásico de programación, pues supone acceder a reglas más po-tentes y flexibles, que se van ajustando automáticamente al entorno a medida que los datos de la muestra van evolucionando y reconociendo variaciones en función de muchísimos factores. Sin embargo, el grado de sofisticación de es-tos algoritmos conlleva una complejidad que, a su vez, incide en la dificultad para su interpretación. El enfoque normativo desconoce la complejidad antes mencionada, gene-rando un marco regulatorio insuficiente y rígido. El presente artículo propone superar dicha visión a través de propuestas flexibles, comprendiendo el rol de la tecnología, reconociendo su valor y estimulando su uso. En definitiva, propuestas que fomenten y fortalezcan la innovación en Chile.
As technology evolves, it will become increasingly frequent for de-cisions such as personnel selection, credit granting or partner selection, to be adopted by machine learning algorithms. Many of these decisions may contain biases that result in discriminatory decisions that affect the subject on whom the decision falls.In order to reveal how such decisions have been made, the idea of transparency in terms of algorithms has been used, implementing it through regula-tions. Such normative or regulatory approach ignores a basic premise: trans-parency is not only a legal problem, but also a technological problem, which becomes relevant when we talk about machine learning algorithms. Machine learning algorithms are the basis for the development of a more sophisticated technology in automating processes than the classic program-ming approach, since it involves access to more powerful and flexible rules, which are automatically adjusted to the environment as the sample data evolve and recognize variations depending on many factors. However, the degree of sophistication of these algorithms leads to a complexity of their models, which generates a difficulty in their interpretation. The regulatory approach does not take into account the complexity mentio-ned above, what generates an insufficient and rigid regulatory framework. This article proposes to exceed this vision through flexible proposals, understan-ding the role of technology, recognizing its value and stimulating its use, this means proposals that promote and strengthen innovation in Chile.
As technology evolves, it will become increasingly frequent for de-cisions such as personnel selection, credit granting or partner selection, to be adopted by machine learning algorithms. Many of these decisions may contain biases that result in discriminatory decisions that affect the subject on whom the decision falls.In order to reveal how such decisions have been made, the idea of transparency in terms of algorithms has been used, implementing it through regula-tions. Such normative or regulatory approach ignores a basic premise: trans-parency is not only a legal problem, but also a technological problem, which becomes relevant when we talk about machine learning algorithms. Machine learning algorithms are the basis for the development of a more sophisticated technology in automating processes than the classic program-ming approach, since it involves access to more powerful and flexible rules, which are automatically adjusted to the environment as the sample data evolve and recognize variations depending on many factors. However, the degree of sophistication of these algorithms leads to a complexity of their models, which generates a difficulty in their interpretation. The regulatory approach does not take into account the complexity mentio-ned above, what generates an insufficient and rigid regulatory framework. This article proposes to exceed this vision through flexible proposals, understan-ding the role of technology, recognizing its value and stimulating its use, this means proposals that promote and strengthen innovation in Chile.