Repository logo
 

Bridging the Gap: Effective Frameworks for Ethical AI Governance - A Systematic Literature Review

Files

Dissertation(886.47 KB)

Date

2026

Authors

Lee, John

Supervisor

Ho, Marcus

Item type

Dissertation

Degree name

Master of Business

Journal Title

Journal ISSN

Volume Title

Publisher

Auckland University of Technology

Abstract

Artificial intelligence (AI) is increasingly embedded in organisational decision-making, reshaping how work is performed, how services are delivered, and how risks are identified and managed. Alongside these opportunities, however, AI introduces material ethical, legal, and reputational risks, particularly in high-impact contexts where automated or model-assisted judgements can affect people’s rights, access to services, and life chances. In response, AI ethics principles and guidelines have proliferated globally; yet comparative reviews highlight substantial variation in scope, terminology, and enforceability, contributing to uneven translation from high-level commitments into operational organisational practice. This persistent “principles-to-practice” gap is widely recognised as a core limitation of contemporary AI ethics discourse, because principles alone rarely provide the procedural specificity, organisational incentives, and accountability mechanisms required to reliably shape decisions under real constraints. This dissertation addresses that gap by synthesising evidence on organisational AI governance through a systematic literature review (SLR), reported in alignment with PRISMA 2020 guidance and informed by evidence-based approaches to management knowledge development. The study examines (1) how organisations implement and adapt governance frameworks to support ethical AI use, and (2) which governance approaches appear most effective for ensuring responsible AI use across organisational contexts. The synthesis identifies recurring patterns through which organisations operationalise ethical aspirations into structures, such as defined roles and oversight forums, processes, for example, risk and impact assessments across the AI lifecycle, and socio-technical practices, such as documentation, auditability, and monitoring. It further shows that governance effectiveness is context-dependent and increasingly shaped by regulatory momentum, including risk-based regimes that formalise differentiated obligations by system risk and use case. Overall, the dissertation concludes that robust AI governance is best understood as an organisational capability rather than a static framework: it requires enforceable accountability, fit-for-purpose operational controls, and continuous adaptation as technologies, organisational incentives, and societal expectations evolve.

Description

Keywords

Source

DOI

Publisher's version

Rights statement

Permanent link

http://hdl.handle.net/10292/21183

Collections

Masters Dissertations