​Winners

Winner: Congratulations to Ryan Kearns, MSc Social Data Science

Thesis title: Quantifying construct validity in large language model evaluations

Ryan says: “I am honoured that my thesis was selected for this award. This project benefitted from the generous support and assistance of so many people, and this achievement is shared with them. I am grateful for the Reasoning with Machines Lab, which provided an excellent community and invaluable feedback at several stages.

Most especially, I thank my advisor, Dr Adam Mahdi, for his encouragement, tireless attention to detail, and continual pushing to turn this thesis into its final product. Thank you also to Jonathan and Andrew for introducing me to this literature and providing their mentorship.

We are privileged to exist at a pivotal inflection point in artificial intelligence—one that urges a deeper understanding of artificial cognition and capabilities. Evaluations are our ‘telescopes’ into this capability space, and I hope that this thesis helps polish the lenses. Thank you to the OII for providing a wonderful interdisciplinary setting for my research, and one I am happy to call home for more years to come through the DPhil.”

Supervisor Dr Adam Mahdi, says: “Ryan’s thesis addresses one of the most important challenges in today’s AI research: how to meaningfully interpret benchmark results for large language models.

By combining insights from social science and computer science, Ryan lays the foundations for a framework that identifies robust, interpretable capabilities rather than relying on potentially misleading headline scores. It was a pleasure to supervise Ryan’s thesis, and he is a thoroughly deserving recipient of the MSc SDS Thesis Prize.”

Winner: Congratulations to Jie Zhao, MSc Social Data Science

Thesis title: Social Crises and the Organized Irresponsibility: Forensic Analysis of Crisis-Time Narrative Control

Jie says: “I came to the OII with questions about power, technology, and whose experiences get to count in our systems, and this thesis is a record of what I was able to answer by pursuing those questions with method and care. It examines how crisis-time narratives form, how public condolences fade out over time after collective traumas, and why some voices become easier to lose than others.

At its core, it reflects the questions I kept returning to at the OII about the opacity and complexity between social actors, and how these can be explained through computational and causal approaches. I am deeply grateful to the OII for the academic freedom that let every curiosity become explorable research question, and for the cohort whose generosity built up the year.

I’m especially thankful to my supervisor, Dr Mohsen Mosleh, for trusting me to take this work seriously and for the insight he brought to every conversation, and to the wider OII community for holding intellectual rigour and moral seriousness in the same place. I leave with deep gratitude for the year, and for everything the OII has made possible for me.”

Supervisor Dr Mohsen Mosleh, says: “This is a richly deserved award. Jie’s thesis stands out for its ambition, care, and intellectual maturity, addressing a difficult empirical case with sensitivity and analytical depth. Supervising her has been a real pleasure, and I’m very proud of what she has accomplished.”

Winner: Congratulations to Maximilian Kroner Dale, MSc Social Science of the Internet

Thesis title: Who should steer AI responses? A survey experiment comparing legitimacy perceptions of participatory versus closed-door AI moderation approaches

Maximilian says: “I am incredibly honoured to receive this award. My research explored how the public views different ways of establishing guardrails for AI chatbot responses. This is a relatively new domain for content moderation, and one that is becoming increasingly politically contested.

Hence, it is a domain where legitimacy in moderation decision-making is particularly important. Empirically, I used a multi-factorial, multi-level survey experiment to measure the causal impact of two different decision-making approaches on participants’ perceptions of legitimacy.

Supervisor Professor Greg Taylor, says: “It was a real pleasure to work with Maximillian on his thesis project. He used a sophisticated experiment to provide deep insights into the perceived legitimacy of approaches to governing AI-generated content. The work is relevant, rigorous, and highly deserving of the Thesis Prize.”

Winner: Congratulations to Sophie Lloyd-Hurwitz, MSc Social Science of the Internet

Thesis title: Evidence-to-Policy Translation in the Social Sciences: A Mixed Methods Analysis of Evidence Selection and Interpretive Fidelity in Social Media & Adolescent Mental Health Policy Recommendations

Sophie says: “I am delighted and honoured to receive this award for my thesis. In my work, I analyse how major science organisations communicate scientific evidence on the effects of social media on adolescent mental health. It is my hope that this research contributes to a more sophisticated understanding of how social science operates in digital policy contexts—and ultimately, to policies that promote the best aspects of the digital world while minimising the worst.

Supervisors Professor Andrew Przybylski and Dr Victoria Nash, say: “Sophie’s thesis cuts straight to the heart of how academic research is used and misused by professional science organizations. By examining the science of social media and adolescent mental health her work identifies evidence synthesis practices in a high-stakes social science domain.

Her thesis provides both an empirical critique and theoretically informed set of concrete recommendations for improvements. It’s a landmark piece of work that should be read by anyone curious about the challenge and promise of evidence-based policy of the internet in the social sciences.”

Highly commended

Highly Commended: Congratulations to Charlie Wang, MSc Social Data Science

Thesis title: Choosing Sides in the Digital Cold War: Alliance Politics, Strategic Hedging, and the Global Battle over Huawei’s 5G Infrastructure

Highly Commended: Congratulations to Yasmeena Khan, MSc Social Science of the Internet

Thesis title: From Manifesto to Movement: Effective Accelerationism and Silicon Valley