• Login
• Admin

Baguley, Kathryn Rebecca (2025) Development and evaluation of a framework to support accountability in Automated Decision-Making (ADM). PhD thesis, University of Nottingham.

PDF (Thesis after corrections) (Thesis - as examined) - Repository staff only until 26 July 2026. Subsequently available to Anyone - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader Available under Licence Creative Commons Attribution. Download (6MB)

Abstract

This PhD uses a mixed methods multidisciplinary approach to assist Small and Medium-sized Organisations (SMOs) in understanding the General Data Protection Regulation (GDPR) and increase their accountability when looking to implement Automated Decision-Making (ADM) technologies. To support SMOs with this task, the PhD develops and evaluates a framework to support accountability from Legal, Ethical and Practical perspectives.

I discuss SMOs' challenges in navigating the GDPR when considering implementing ADM technologies. I also present empirical findings from publicly available Ethical Documents (ED) for ADM and use these findings, alongside those of the literature reviews, to design and construct the first proof-of-concept (POC) tool, which I envisage as part of an Automated Decision-Making Impact Assessment (ADMIA). The concept of the ADMIA overall is to be a modular tool-based framework solution to help SMOs understand accountability and decide whether to implement ADM. I create a POC to help users understand the EUGDPR Article 22 (ADM22). I then undertake a study to obtain feedback on the POC to provide insights for improvement and additions for future iterations.

ADM is transforming the world in ways we do not always appreciate or see. While using ADM or algorithms to make decisions is not new, combining big data and high computing power means there is a driving need for better understanding and protection for all. Organisations need tools to support them in their journey to understanding and navigating complex legislation and enabling a robust manner to record and account for their decisions. SMOs may face similar legislative complexity and accountability challenges whether using ADM or not; however, ADM tends to increase complexity and scale, exacerbating the likelihood of matters such as discrimination and bias.

This work predominantly focuses on the legal definitions, such as those in the UK and EU GDPR(s). The lack of universal definitions contributes to the challenges with technology continually progressing alongside experts across disciplines failing to reach any agreement.

This PhD highlights the need for more structured documentation for SMOs to enable a better understanding of ADM and accountability. I designed the POC to appeal to SMOs on two levels. Firstly, this is a tool to help guide them through complex legislation, therefore helping with the necessity of GDPR legal compliance. Using this tool should support users and the SMO with a means to understand their GDPR legal obligations at a level which suits their experience. However, should the SMO discover no GDPR compliance concerns within their intended project, the POC also guides best practices to increase accountability over the decision-making process.

The PhD produces a POC section of the ADMIA as the main contribution to academic knowledge and as a proposed practical solution to assist SMOs in attaining and maintaining accountability when implementing ADM technologies into their organisations. Insights following the design and construction of the POC show that tools are a practical solution to assist SMOs with accountability when implementing ADM technologies.

Item Type:	Thesis (University of Nottingham only) (PhD)
Supervisors:	Hyde, Richard Fischer, Joel
Keywords:	GDPR; AI; Automated Decision-Making; Accountability; Tools
Subjects:	K Law > KJ Europe
Faculties/Schools:	UK Campuses > Faculty of Social Sciences, Law and Education > School of Law
Item ID:	80569
Depositing User:	Baguley, Kathryn
Date Deposited:	26 Jul 2025 04:40
Last Modified:	26 Jul 2025 04:40
URI:	https://eprints.nottingham.ac.uk/id/eprint/80569

Actions (Archive Staff Only)

Edit View