Psychometrics, profiles and bias: The case for facial recognition

Psychometric studies allow measurements based on vector correlations that produce predictions of human behavioral traits. With increasing digitization, psychometrics has been used to design profiles of individuals through various mechanisms: Big Data, Machine Learning, among others. Far from being n...

Descripción completa

Detalles Bibliográficos
Autor: Gómez, Juan Camilo
Tipo de recurso: artículo
Estado:Versión publicada
Fecha de publicación:2021
País:Uruguay
Institución:Universidad ORT Uruguay
Repositorio:RAD
Idioma:español
OAI Identifier:oai:rad.ort.edu.uy:20.500.11968/4326
Acceso en línea:https://revistas.ort.edu.uy/inmediaciones-de-la-comunicacion/article/view/3156
https://hdl.handle.net/20.500.11968/4326
http://hdl.handle.net/20.500.11968/4326
Access Level:acceso abierto
Palabra clave:psychometrics
profile-dividum
facial recognition
bias
psicometría
perfil-dividuo
reconocimiento facial
sesgos
psicometria
reconhecimento facial
vieses
Descripción
Sumario:Psychometric studies allow measurements based on vector correlations that produce predictions of human behavioral traits. With increasing digitization, psychometrics has been used to design profiles of individuals through various mechanisms: Big Data, Machine Learning, among others. Far from being neutral, these profiles are based on methodologies which produce biases that affect the profiled individuals. This article proposes that psychometrics elaborates a dividual-profile that generates a reduction of individuals and produces forms that augur the way in which they should behave. Here the conjecture is exemplified from the analysis of a case of psychometric measurement based on facial recognition.