Algorithmic bias is a widely recognised but still under-researched problem. Design choices in AI systems, such as the training data used in natural language models and representations, may reflect and reinforce existing social and cultural inequalities, regardless of the intentions of developers. In H&SC, the potential of AI has been widely celebrated, but actual adoption and implementation by the NHS is still quite sparse, making this an ideal time to identify, mitigate and share learning about any algorithmic gender bias. Incomplete data in relation to gender and other variables, collected by GPs, can result in biases that have significant impact on accuracy and fairness of AI algorithms for example. In line with recent developments in the field, the proposed interdisciplinary social science and computer science project will research how algorithmic gender bias occurs and may result in risks to equality of access and outcomes in H&SC provision and how it can best be mitigated against.
The PhD student will receive training to facilitate the use of a combination of qualitative social sciences methods such as qualitative interviewing, quantitative social sciences methods such as analysis of statistical data, and computer science analytical skills, such as data cleaning. The student will collaborate closely with H&SC providers and AI
system developers to improve approaches to system design, implementation and bias-testing.
The key overall outcome of the project will be new knowledge published in the PhD and in peer-reviewed journals about algorithmic gender bias, how to understand and address it, in order to facilitate equality of access and outcome through AI-based systems in H&SC.
The University of Brighton has internationally recognised strengths in AI research in social science, arts and humanities and computer science contexts, spanning research fields such as data analytics, natural language processing, human-computer interaction and systems security, computer graphics, dialogue and narrative generation, and novel user experiences including augmented/virtual/mixed reality. The student will be formally located in the School of Applied Social Sciences but have access to outstanding cross-disciplinary research environments through participation in both the Digital Media Cultures and Applied Data Analytics research centres, enabling them to develop the necessary skills and competences in social framing of technologies, computational technologies, economic imperatives, cultural effects, ethics and governance of artificial intelligence for future societies.