Trustworthy machine learning: mitigating bias and promoting fairness in automated decision systems
Year: 2023
Type: dissertation
Abstract: The widespread adoption of Artificial Intelligence (AI) and Machine Learning (ML) algorithms in recent years across many domains has led to an increased reliance on automated decision making. While these algorithms have shown tremendous promise in improving decision-making efficiency and accuracy, they are not immune to errors and biases. Consequently, there is a growing concern about the trustworthiness of automated decision-making systems (ADMS). Trustworthiness refers to the degree to which an AD-MS can be relied upon to produce accurate, fair, robust, transparent, inclusive, and empowering results. Ensuring the trustworthiness of AD-MS is crucial in several domains, among many others, healthcare, finance, criminal justice, and human resources. For instance, biased or inaccurate automated credit scoring systems can result in unfair denial of loans to certain individuals, while biased recruitment systems can perpetuate discrimination in the workplace. Consequently, there is a need for tools and approaches to improve the trustworthiness of AD-MS. This project aims to explore the challenges and opportunities in achieving trustworthy automated decision-making. Specifically, it seeks to investigate the limitations of machine learning algorithms, the importance of trustworthiness, and the tools and approaches for improving trustworthiness. The project also aims to examine the ethical, legal, and social implications of AD-MS and present case studies that illustrate the practical application of the proposed tools and approaches. The project concludes by discussing future directions for research in this field.
Author: Wolters, Stephan
Institution: Universidad Politécnica de Madrid
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Keywords: trustworthy machine learning, automated decision systems, algorithmic bias, fairness, algorithmic discrimination, algorithmic transparency, algorithmic accountability, ethical implications of AI, legal implications of AI, societal impact of AI, credit scoring, recruitment system, algorithmic robustness
Open Access status: green
OpenAlex ID: W7025174994
