Publicação

Anthropocentric AI for EU consumer lending: ‘Hacking’ opacity and discrimination ‘sins’ powered by creditworthiness machines

Ver documento

Detalhes bibliográficos
Resumo:Incorporating AI-based decision-making into consumer credit assessment under the framework of Consumer Law enhances regulatory compliance. This paper outlines a Multi-Agent Systems (MAS) to implementing Art. 18(6)(8)(9) of the EU 2023/2225 Directive, dated 18 October. In pursuit of this goal, we propose a legal framework emphasizing the necessity of hybrid oversight in AI-based consumer scoring. This study aims to improve transparency and fairness through the implementation of an Explainable Agent-based layer. Overall, this research introduces the concept of Machine-Centred Anthropocentrism. It acknowledges that, after the training, validation and testing stages, credit analysts no longer have complete psychological control over the data-driven entities programmers have given birth.
Autores principais:Rebelo, Diogo Morgado
Outros Autores:Andrade, Francisco Carneiro Pacheco; Novais, Paulo
Assunto:Art. 18 of the EU 2023/225 Directive Creditworthiness Assessment Multi-Agent-Centred Anthropocentrism
Ano:2025
País:Portugal
Tipo de documento:comunicação em conferência
Tipo de acesso:acesso aberto
Instituição associada:Universidade do Minho
Idioma:inglês
Origem:RepositóriUM - Universidade do Minho
Descrição
Resumo:Incorporating AI-based decision-making into consumer credit assessment under the framework of Consumer Law enhances regulatory compliance. This paper outlines a Multi-Agent Systems (MAS) to implementing Art. 18(6)(8)(9) of the EU 2023/2225 Directive, dated 18 October. In pursuit of this goal, we propose a legal framework emphasizing the necessity of hybrid oversight in AI-based consumer scoring. This study aims to improve transparency and fairness through the implementation of an Explainable Agent-based layer. Overall, this research introduces the concept of Machine-Centred Anthropocentrism. It acknowledges that, after the training, validation and testing stages, credit analysts no longer have complete psychological control over the data-driven entities programmers have given birth.