Please use this identifier to cite or link to this item: https://hdl.handle.net/2445/224023
Biaix algorítmic a la IA: conceptes clau, implicacions i solucions
Journal Title
Director/Tutor
Journal ISSN
Volume Title
Related resource
Abstract
[cat] Aquest recurs educatiu obert (REA) forma part del projecte europeu GEDIS – Gender Diversity in Information Science: Challenges in Higher Education, cofinançat per la Unió Europea. El material, titulat Biaix algorítmic en la IA: conceptes clau, implicacions i solucions, és una traducció al català del recurs original Algorithmic Bias in AI: Key Concepts, Implications, and Solutions (Karmil, Boskovic i Kartašová, 2025). La traducció ha estat realitzada per Claudia San-José i Juan-José Boté-Vericad.
El recurs introdueix els conceptes fonamentals relacionats amb el biaix algorítmic en la intel·ligència artificial (IA), les seves causes i les repercussions socials i ètiques associades. Explica com els sistemes d’IA, en analitzar dades històriques i actuar d’acord amb patrons estadístics, poden reproduir o amplificar desigualtats preexistents de gènere, raça o classe. A través d’exemples reals en àmbits com la salut, el dret o els recursos humans, el material mostra com els algorismes poden produir resultats discriminatoris o injustos si no s’apliquen amb una supervisió adequada.
El recurs posa èmfasi en la necessitat de supervisió humana i transparència en totes les etapes del disseny i l’ús de la IA. S’hi descriuen diverses vies de mitigació, com ara:
la diversificació i representativitat de les dades d’entrenament;
l’establiment d’auditories periòdiques d’exactitud i imparcialitat;
la definició de directrius ètiques i reguladores;
i la incorporació d’eines de IA per detectar possibles biaixos.
Aquest OER també promou una lectura crítica i reflexiva de la tecnologia: recorda que la IA no raona com els humans, sinó que prediu estadísticament en funció de les dades disponibles. D’aquesta manera, fomenta la capacitat de l’estudiantat i del personal docent per identificar biaixos i desenvolupar una consciència informada sobre l’impacte de la IA en la societat. #GEDIS #SummerSchoolBarcelona
[eng] This Open Educational Resource (OER) is part of the European GEDIS Project – Gender Diversity in Information Science: Challenges in Higher Education, co-funded by the European Union. The material, titled Algorithmic Bias in AI: Key Concepts, Implications, and Solutions, is the Catalan translation of the original resource Algorithmic Bias in AI: Key Concepts, Implications, and Solutions (Karmil, Boskovic, and Kartašová, 2025). The translation was carried out by Claudia San-José and Juan-José Boté-Vericad. The resource introduces the fundamental concepts related to algorithmic bias in artificial intelligence (AI), its causes, and its associated social and ethical implications. It explains how AI systems, by analysing historical data and acting according to statistical patterns, can reproduce or amplify pre-existing inequalities related to gender, race, or class. Through real-world examples in areas such as healthcare, law, and human resources, the material illustrates how algorithms may produce discriminatory or unfair outcomes when not implemented under proper supervision. The resource emphasises the need for human oversight and transparency at all stages of AI design and deployment. It outlines several mitigation strategies, including: diversifying and ensuring representativeness in training datasets; establishing periodic audits for accuracy and fairness; defining ethical and regulatory guidelines; and incorporating AI tools to detect potential biases. This OER also promotes a critical and reflective approach to technology, reminding readers that AI does not reason like humans but rather makes statistical predictions based on available data. In doing so, it fosters the ability of students and educators to identify biases and develop informed awareness of AI’s impact on society. #GEDIS #SummerSchoolBarcelona
[eng] This Open Educational Resource (OER) is part of the European GEDIS Project – Gender Diversity in Information Science: Challenges in Higher Education, co-funded by the European Union. The material, titled Algorithmic Bias in AI: Key Concepts, Implications, and Solutions, is the Catalan translation of the original resource Algorithmic Bias in AI: Key Concepts, Implications, and Solutions (Karmil, Boskovic, and Kartašová, 2025). The translation was carried out by Claudia San-José and Juan-José Boté-Vericad. The resource introduces the fundamental concepts related to algorithmic bias in artificial intelligence (AI), its causes, and its associated social and ethical implications. It explains how AI systems, by analysing historical data and acting according to statistical patterns, can reproduce or amplify pre-existing inequalities related to gender, race, or class. Through real-world examples in areas such as healthcare, law, and human resources, the material illustrates how algorithms may produce discriminatory or unfair outcomes when not implemented under proper supervision. The resource emphasises the need for human oversight and transparency at all stages of AI design and deployment. It outlines several mitigation strategies, including: diversifying and ensuring representativeness in training datasets; establishing periodic audits for accuracy and fairness; defining ethical and regulatory guidelines; and incorporating AI tools to detect potential biases. This OER also promotes a critical and reflective approach to technology, reminding readers that AI does not reason like humans but rather makes statistical predictions based on available data. In doing so, it fosters the ability of students and educators to identify biases and develop informed awareness of AI’s impact on society. #GEDIS #SummerSchoolBarcelona
Description
This is an OER produced within the GEDIS Project – Gender Diversity in Information Science: Challenges in Higher Education.
Subject (English)
Citation
San José, Claudia, Boté-Vericad, Juan-José. 2025. "Biaix algorítmic a la IA: conceptes clau, implicacions i solucions." Zenodo. https://doi.org/10.5281/zenodo.17306646.
Collections
Citation
BOTÉ-VERICAD, Juan-José and SAN JOSÉ, Claudia. Biaix algorítmic a la IA: conceptes clau, implicacions i solucions. [consulted: 11 of June of 2026]. Available at: https://hdl.handle.net/2445/224023