Participatory Algorithmic Justice: A multi-sited ethnography to advance algorithmic justice through participatory design

Descripción del proyecto

Dar voz a los afectados por la inteligencia artificial

La inteligencia artificial (IA) promete revolucionar nuestro mundo, pero también causa daños. Desde los prejuicios raciales y de género hasta las detenciones y la vigilancia injustas, los sistemas de IA suelen acentuar las desigualdades existentes. Las tecnologías pueden discriminar a los grupos marginados y dar lugar a un trato injusto. La mayoría de las personas más afectadas tienen poco que decir sobre cómo se desarrolla o utiliza la IA. En este contexto, el proyecto PARTIALJUSTICE, financiado por el Consejo Europeo de Investigación, pretende cambiar esta situación poniendo las voces de estas comunidades en el centro del desarrollo de la IA. Se centra en la justicia algorítmica participativa, un nuevo enfoque que implica a los grupos marginados en la elaboración de soluciones a los efectos nocivos de la IA. A través de trabajos de campo y talleres, el proyecto ayuda a identificar problemas y diseña intervenciones.

Objetivo

AI comes with great promises, but more critically, it comes with demonstrated harms, from racial and gender bias, to discrimination, wrongful arrests, defamation, surveillance, or the extractive labor and practices required for its creation. I contend that the most concerning aspect of AI is the exacerbation of existing structural inequalities and creation of new ones. Two prevailing responses have emerged to the social and ethical problems of AI: ethicists and social scientists have focused largely on principle-based approaches, and developers have focused on technical fixes for addressing problems in AI models. This project speaks to an urgent gap between these approaches: it centers the experiences of marginalized communities and develops interventions into AI concerns that are designed by and for those communities.

This project develops the novel approach of participatory algorithmic justice. Participatory algorithmic justice defines a concept and standards of practice for collaborative research to better understand who and what AI harms, but also how these harms should be redressed. The project investigates how economic, cultural, and political harms from AI are experienced by structurally marginalized groups through multi-sited, intersectional ethnographic fieldwork. This fieldwork informs participatory design workshops to develop specific interventions into problems identified by research collaborators. This research is brought to life through an innovative combination of graphic storytelling and a public-facing mapping platform. Participatory algorithmic justice brings the voices, priorities, and concerns of those affected by AI to the forefront of debates over what kinds of AI people want to live with. Tackling a critical problem of global technology justice, this project is a crucial intervention for communities, researchers, and developers to redress AI harms.