An interdisciplinary study by the University of Bristol uncovering the socio-economic, literacy, and structural factors shaping the future of AI research.
"Getting Started in AI" training sessions delivered at the University of Bristol.
Our research utilised a robust mixed-methods approach to map the transition from undergraduate to doctoral AI studies. We developed a custom Natural Language Processing (NLP) pipeline to analyse over 2,600 PhD recruitment listings, identifying linguistic and structural barriers in funding and eligibility.
Parallel to this data-driven analysis, we collaborated with the Jean Golding Institute (JGI) to deliver "Getting Started in AI" training sessions. These workshops served as an intervention for students from non-technical backgrounds, allowing us to track shifts in self-efficacy and study intentions.
2,643 PhD opportunities scraped and analyzed for structural exclusion.
JGI-led workshops designed to demystify AI for underrepresented groups.
45-60 minute sessions exploring student trust and perceived barriers.
Addressing the fundamental challenges in AI doctoral accessibility and institutional reform.
The AI talent shortage is a bottleneck for global innovation. By expanding the pipeline, we ensure the most capable minds shape the next era of technology.
Inclusion is critical for preventing bias. Diverse researchers bring the perspectives necessary to build systems that are fair and representative of all society.
Our findings reveal systemic "False Hope" barriers. We provide evidence-based recommendations to help universities reform their recruitment pipelines.
Explore each research section directly.
Theoretical framework and identified barriers.
Go to PageNLP pipeline and research design details.
Go to PageResults from data analysis and interviews.
Go to PageAccess datasets, reports, and toolkits.
Go to PageMeet the interdisciplinary researchers.
Go to PageInvestigating how gender, race, and financial background form "invisible gates" in the PhD application pipeline.
Utilising the AILIT-S framework to distinguish technical ability from the psychological confidence gap.
Auditing institutional policy misalignment between advertised inclusivity and restrictive funding caps.
Our robust sequential mixed-methods design moves from theoretical scoping to deep analytical interventions.
Identified core dimensions of exclusion: socio-economic, literacy, and structural.
Collected 2,643 live PhD listings from FindAPhD.com for analysis.
JGI-led "Getting Started in AI" workshops for non-technical students.
Longitudinal Pre/Post tracking using the AILIT-S framework.
Semi-structured sessions probing personal narratives and trust.
Triangulation of quantitative stats and qualitative reflexive thematic analysis.
"While 79.2% of listings claim global openness, recruitment caps create a 'False Hope' barrier for international talent."
"Lower socio-economic students report lower self-efficacy despite academic parity with their peers."
Open science repository featuring all research outputs and pedagogical tools.
Final project outputs and executive summaries.
Syllabi and slides from our training workshops.
AILIT-S survey items and interview guides.
Web visualisation scripts and interactive data code.
Jana Alloush
Erin Brady
Rosalba Castiglione
Weilin He
Sven Hollowell
Adrianna Jezierska
Pau Erola
Emily Wride
Prof. Flavia De Luca