Julian Hazell
Former MSc Student
Julian is interested in researching how to ensure that transformative AI benefits humanity. He worked as a research assistant for the policy team at the Centre for the Governance of AI.
Introducing the 2023 MSc Thesis Prize Winners
We are delighted to announce the winners of the 2023 OII MSc Thesis Prizes. Two prizes were awarded for the best theses from the MSc in Social Science of the Internet and the MSc in Social Data Science, plus two ‘Highly Commended’ awards.
Winner: Congratulations to Julian Hazell, MSc Social Science Of The Internet (Part-Time)
Thesis title: The Sentinel Model: A Study of Defensive AI Systems and Misuse Risks from Advanced AI
Supervisor: Professor Viktor Mayer-Schoenberger, Oxford Internet Institute
Julian said: “I’m honoured to receive this award. Having the opportunity to explore the fascinating topic of defensive AI systems was a formative experience, and I’m grateful to the various faculty members at the OII — especially my thesis supervisor Viktor — who encouraged me to follow my interests when choosing a thesis topic. I would also like to extend a sincere thank you to my former colleagues and peers at GovAI, the OII, and the broader AI governance community, who have all taught me so much. I hope that this thesis will inspire further research into how defensive applications of AI systems can improve society’s resilience against various risks posed by the development of transformative AI systems
In congratulating Julian, Professor Mayer-Schoenberger said, “Everyone talks about AI these days – what miracles it can do, and how dangerous it might be. Rather than such pontifications, what’s really needed are pragmatic inquiries about whether and to what extent AI can help us address some of the challenges it poses. Julian Hazell’s thesis does exactly that. And by showing what is and isn’t realistic at the current state, Julian’s work helps us make sense of the world of AI we encounter. Congratulations, Julian!”
Highly Commended: Katherine Crowley
Thesis title: Untraceable Digital Advertising in U.S. Politics: The Case for Regulatory Reform
Winner: Congratulations to Harry Mayne, MSc Social Data Science
Thesis title: Unsupervised Learning Approaches to Intensive Care Reform: Opportunities and Challenges
Supervisor: Director of Graduate Studies, Departmental Research Lecturer, Dr Adam Mahdi, Oxford Internet Institute
Harry said; “I’m honoured that my thesis has been chosen for this award. The project turned out to be a really exciting piece of research to work on and I was delighted with the outcome of it. The thesis itself develops a framework for how unsupervised machine learning might be used to improve the efficiency of intensive care units without negatively affecting the quality of care that patients receive. A preprint focusing on the first part of the framework will be available on arXiv soon for those interested in finding out more. Working on the thesis was undoubtedly a demanding process, but incredibly rewarding once I started looking at the implications of the results. I’m extremely thankful to all those who supported the project, with special thanks to Dr Adam Mahdi, whose supervision was invaluable to its success, and Dr Guy Parsons, for providing medical insight throughout the process. The Social Data Science MSc was a brilliant course and I met so many talented people who I have no doubt are going to go on to do amazing things. From working with this year’s cohort, it seems this was no exception, so I’m looking forward to seeing what research they produce”.
In congratulating Harry, Dr Mahdi said, “I am immensely delighted of Harry’s achievement in receiving the 2022-23 OII Thesis Prize for Social Data Science. Harry’s dedication and insights, from the beginning of the project, was exceptional. Through the use of unsupervised machine learning, Harry demonstrated how one could potentially identify subgroups of intensive care patients with shared levels of medical resource needs. Consequently, he discussed how ICUs could be restructured into smaller sub-units, each catering to a specific group. Harry’s work exemplifies the caliber of research produced by the Social Data Science program, and I have no doubt that his insights will continue to shape the future of data science. My congratulations to Harry on this well-deserved recognition.”
Highly Commended: Khyati Khandelwal
Thesis title: Casteist but not Racist? Towards Quantifying and Mitigating Bias in Large Language Models
Congratulations once again to all our winners!