ESSAYS
Two papers on artificial intelligence: how a game makes decisions, and how the data behind AI should be protected.
01 / SUMMER SCHOOL PAPER
Building a Tic-Tac-Toe Game with Minimax
Proudly published by Fitzwilliam College, University of Cambridge.
2025 Fitzwilliam College Winter and Summer School Journal · Roger Yeoh · pp. 52–54
I wrote this paper alongside my Minimax project at the FitzEd summer school. It follows the game from a two-player Python program to a Tkinter interface, then to an AI that searches the full game tree.
The part I wanted to understand was how a board position becomes a decision. I explain how recursion explores possible moves, how final positions are scored, and how backtracking chooses a move. The paper compares a simple opponent that checks immediate wins and blocks with a version that searches to the end of the game.
Full search makes optimal play possible in Tic-Tac-Toe, but my implementation does not prune branches. I discuss that limitation: searching every possibility becomes expensive as the game grows.
Opens the official journal at page 52 in a new tab.
02 / EPQ RESEARCH PAPER
Research on artificial intelligence data protection in China
I started with the pace of AI development and the possibility of maliciously polluted training data. At the time, I was watching China’s AI industry work to catch up and wondering whether the protections around its data were keeping pace.
The paper focuses on those protections. I examine China’s AI development strategy, its data protection laws, and the way different departments enforce them. I also compare the framework with principles in the EU’s GDPR.
My central argument is that AI development needs data governance that works in practice. I focus on clearer responsibilities between regulators, protection matched to the importance of the data, safeguards for cross-border transfers, and more understandable privacy policies.
“Enterprises need to explore data sharing norms and apply privacy enhancement technologies to balance data value and privacy protection.”
From my conclusion
I argue for giving people more control over how their information is collected, used and shared. Effective protection also matters for the industry: without it, data misuse can undermine trust in the technology and make international cooperation harder.