Publicação
Bias in Artificial Intelligence: Exploring Its Role in Institutional Discrimination and Strategies for Mitigation
| datacite.subject.fos | Ciências Naturais::Ciências da Computação e da Informação | pt_PT |
| dc.contributor.advisor | Montargil, Filipe José de Oliveira Frescata e Marques | |
| dc.contributor.author | Oberhauser, Hubert Josef | |
| dc.date.accessioned | 2025-02-24T15:00:24Z | |
| dc.date.available | 2025-02-24T15:00:24Z | |
| dc.date.issued | 2025-02-13 | |
| dc.description | Dissertation presented as the partial requirement for obtaining a Master's degree in Data Science and Advanced Analytics, specialization in Data Science | pt_PT |
| dc.description.abstract | Artificial Intelligence, especially through machine learning, has become ubiquitous in everyday life, influencing both individual choices and institutional decision-making processes. Despite its potential to bring societal benefits, AI has faced considerable criticism because it perpetuates biases contributing to institutional discrimination. This thesis explores how biases come into AI systems through imbalance in data, algorithmic design, or user interaction, and how they reinforce systemic inequalities in critical domains such as recruitment, justice, and finance. The thesis is based on an extensive literature review, discussing both the technical foundations of AI and the notion of institutional discrimination defined under the EU and German frameworks as systemic inequalities that are buried within institutional practices. It showcases real-world impacts: the biased algorithms, including Amazon's hiring tool, the COMPAS criminal justice system, and discriminatory credit scoring mechanisms, through which AI systems replicate and amplify historical inequities. These challenges are addressed by this research, which assesses the strategies for mitigating AI bias through technical, regulatory, and ethical approaches. Technical interventions include fairness-aware algorithms and improved data practices, while regulatory measures, such as the EU AI Act and GDPR, enforce accountability and transparency. Ethical frameworks, including the OECD Principles on AI and EU Ethics Guidelines for Trustworthy AI, emphasize inclusivity and fairness in AI governance. By integrating these perspectives, actionable strategies to reduce bias and advance equity in AI systems are provided. It reflects a deep understanding of the power of interdisciplinary collaboration in AI for social progress, fairness, and accountability in its deployment processes. | pt_PT |
| dc.identifier.tid | 203921623 | |
| dc.identifier.uri | http://hdl.handle.net/10362/179666 | |
| dc.language.iso | eng | pt_PT |
| dc.rights.uri | http://creativecommons.org/licenses/by/4.0/ | pt_PT |
| dc.subject | Artificial Intelligence | pt_PT |
| dc.subject | Machine Learning | pt_PT |
| dc.subject | Bias | pt_PT |
| dc.subject | Discrimination | pt_PT |
| dc.subject | Mitigation | pt_PT |
| dc.subject | SDG 9 - Industry, innovation and infrastructure | pt_PT |
| dc.subject | SDG 10 - Reduced inequalities | pt_PT |
| dc.subject | SDG 16 - Peace, justice and strong institutions | pt_PT |
| dc.title | Bias in Artificial Intelligence: Exploring Its Role in Institutional Discrimination and Strategies for Mitigation | pt_PT |
| dc.type | master thesis | |
| dspace.entity.type | Publication | |
| rcaap.rights | openAccess | pt_PT |
| rcaap.type | masterThesis | pt_PT |
| thesis.degree.name | Mestrado em Ciência de Dados e Métodos Analíticos Avançados, especialização em Ciência de Dados | pt_PT |