Background
Medical school applicants from Widening Participation (WP) backgrounds continue to face systemic barriers to effective interview preparation. These barriers disproportionately affect students from low-income households, state schools, and underrepresented ethnic groups. Interviews, particularly the Multiple Mini Interview (MMI) format used by over 75 percent of UK medical schools, have become the decisive stage in determining entry. Yet, access to high-quality, structured interview support remains highly stratified.
To address this challenge, The Aspiring Medics developed AVA, an AI-powered, voice-interactive platform that delivers equitable and scalable interview preparation. AVA simulates real-time MMI stations and provides structured feedback across four key domains: content, insight, structure, and communication. This study evaluates AVA's impact on self-efficacy and interview performance and examines whether outcomes are equitable across demographic groups.
Methods
A longitudinal observational study was conducted with 185 Year 13 students from areas with historically low higher education participation (POLAR1 and POLAR2). Participants were provided with AVA access over an eight-week period. Based on engagement data, students were categorised into Low, Medium, and High Usage groups.
Performance was measured using AVA score quartiles at Weeks 4 and 8. Chi-square tests assessed the relationship between usage and performance. Self-efficacy was measured at Weeks 0, 4, and 8 using a validated five-item scale that assessed confidence in preparation, understanding complex questions, handling difficult questions, recalling key information, and likelihood of success. Paired t-tests and MANOVA were used to assess self-efficacy changes, and Kruskal–Wallis tests evaluated variations by ethnicity and geography.
Results
Overall, self-efficacy improved significantly across the study period. Mean scores increased by +0.38 at Week 4 (p < 0.001) and by +0.77 at Week 8 (p < 0.001), with a medium-to-large effect size (Cohen’s d = 0.70). The most significant gains were in students' confidence with handling difficult questions, understanding complexity, and anticipating success in interviews.
Higher AVA engagement was significantly associated with improved performance at both checkpoints. Chi-square analysis showed clear correlations between usage and success: Week 4 (χ² = 16.813, p = 0.010) and Week 8 (χ² = 15.686, p = 0.016). MANOVA confirmed that usage group had a statistically significant effect on self-efficacy (p = 0.004). Improvements were consistent across ethnic groups (p = 0.461) and geographic regions (p = 0.261), indicating equitable outcomes.
Conclusion
This study provides the first UK-based evidence that an AI-powered interview tool can enhance both self-efficacy and performance for WP students applying to medical school. AVA significantly improved confidence and offer outcomes, with equitable benefits across ethnic and geographic backgrounds. These findings support the use of AI to address structural inequalities in admissions. Further research should explore the long-term impact of AVA on medical school entry and progression.
View the presentation in full here:
Ben-Tarifite, Y. (2025, October 24). AI-Powered Interview Preparation for Widening Participation Students: Evaluating the Impact of AVA, the UK's First AI Interview Platform. AIEOU Inaugural Conference, University of Oxford. Zenodo. https://doi.org/10.5281/zenodo.17537406
The content expressed here is that of the author(s) and does not necessarily reflect the position of the website owner. All content provided is shared in the spirit of knowledge exchange with our AIEOU community of practice. The author(s) retains full ownership of the content, and the website owner is not responsible for any errors or omissions, nor for the ongoing availability of this information. If you wish to share or use any content you have read here, please ensure to cite the author appropriately. Thank you for respecting the author's intellectual property.
