Ethics and transparency for detection of gender bias in algorithms

Abstract

The algorithms' growing importance shows the discrimination registered, especially on gender and minority groups, besides the need of transparency in the application of these formulas against the corporations' opacity. Despite these biases, the making decision on almost all the knowledge fields, as well as the social, political and economic activities, leans on algorithms because of the blind trust in computer processing and the technological imaginary about their ability to eliminate the error and the bias. The Search Engine Manipulation Effect (SEME) (Epstein y Robertson, 2015) shows very clear effects on voting behavior. Caliskan y Bryson (2017) have also detected the reproduction of gender and ethnic biases when working on already biased data, which lead to a very important statistical deviations in Big.