I've published the following two research papers:

Implementing and Evaluating Deep Models for Image Generation: Performance and Ethical Considerations

Under the supervision of Oxford professor Dr. Joe Pitt-Francis, I wrote a cross-discipinary evaluation of Deep Generative AI models, emphasizing the ethically dubious features of these models with regards to privacy. I argue that we ought to embed ethical considerations into the very design of the models, giving constructive ideas for how to do so.

Ant-Run is NP-Complete

With a friend, I proved that the old-school video game Ant-Run is NP-complete. This involved translating abstract game rules into formal computational problems and proving their classification within complexity theory. In the future, I plan to explore how the principles of computational complexity can be applied to other interactive systems and design problems, including VR or interactive simulations.

[Updated early June] I'm also currently working on the following research paper with colleagues at the Oxford Internet Institute:

The Case for Co-Created AI Governance for the Global Majority: Motivations, Benefits, and Recommendations

A lack of AI governance mechanisms in the Global South, combined with regulatory fragmentation, has enabled or exacerbated various forms of AI-driven inequalities and risks. Therefore, I propose that collective, co created AI Governance initiatives rooted in South-South solidarity can reduce certain forms of extraction and exploitation, give GS stakeholders a stronger presence in global AI Governance conversations, and help protect digital sovereignty.