Masterarbeit
Managing AI bias in Web Systems
Completion
2025/03
Research Area
Students
Jannatul Ferdous
Advisers
Dr.-Ing. Sebastian Heil
Description
Artificial Intelligence (AI) has become an integral component of web-based systems, optimizing decision-making and enhancing user experiences. However, the presence of bias in AI systems poses significant ethical, legal, and social challenges, necessitating robust frameworks to ensure fairness, accountability, and compliance with regulatory standards such as the EU AI Act. Software providers need to able to manage biases just as they manage risks related to cybersecurity through systematic approaches for identification and mitigation.
This thesis aims at designing and implementing an approach to systematically describe, identify, and notify architects and developers about biases and mitigations in web-based AI systems. A suitable solution needs to comprise of a uniform machine-readable model for representing AI biases and a supporting web-based platform to capture and query known biases similar to the CVE infrastructure. To demonstrate the platform, it needs to be integrated with an existing modeling approach for web systems so that architects can be notified about potential biases in their systems.
The objective of this thesis is the creation of a solution or the combination of existing techniques to solve the problem of managing AI bias in web-based AI systems through a web-based platform as described above. This comprises the analysis of the state of the art of AI biases and cybersecurity risk management as well as the demonstration of the solution by implementation and a suitable experimental evaluation focusing on the applicability of the approach with known biases from literature and the usability and usefulness of the platform.
