Assessment Frameworks for Trustworthy AI: A Systematic Literature Review

  • Typ: Thesis
  • Zielgruppe: Bachelor
  • Dozent:

    Julia Gutschow

Problem Description

As Artificial Intelligence (AI) systems become increasingly integrated into high-impact areas such as healthcare, finance, hiring, and public services, ensuring their trustworthiness has become a global priority. Trustworthy AI refers to artificial intelligence systems that operate in a reliable, transparent, and ethical manner, addressing concerns such as fairness, accountability, and security. Various frameworks (e.g., Z-Inspection®, NIST AI Risk Management Framework, HLEG Trustworthy AI Guidelines) have been proposed to assess and support the development of AI systems that meet these standards. However, these frameworks differ in scope, approach, and applicability, making it unclear how they compare and how they can be effectively implemented.

Goal of the Thesis

This thesis aims to conduct a systematic literature review of assessment frameworks for Trustworthy AI, analyzing their similarities, differences, and practical implementation. The student will explore how these frameworks define and assess Trustworthy AI, identifying key aspects that emerge from the literature and synthesizing their strengths, limitations, and gaps. The final output will be a comparative report, providing a structured overview of AI assessment methodologies and offering recommendations for improving AI evaluation processes.

Requirements

  • Interest in Trustworthy AI and assessment frameworks
  • Proficiency in conducting a systematic literature review or willingness to learn it independently
  • Ability to summarize and synthesize findings systematically

Sources