Philos Kim is a rising sophomore at Yale studying Mathematics.
As a JURIST Digital Scholar, Philos will examine the extent of algorithmic bias in deep learning models by using JURIST articles as testing data. In particular, he will design sentiment classification networks with both convolutional and recurrent architectures in PyTorch, using Captum to interpret and visualize their parameters and results. He also plans to use statistical methods to draw conclusions about the significance of his results, as well as to compare the performance and bias of complex deep learning models to simpler statistical ones. Ultimately, he wants to draw conclusions about the extent of demographic and ideological bias in cutting-edge artificial intelligence models, and how that manifests when legal news and commentary articles are the test data. Although his project is on the technical side, Philos wants to make the tools he creates and the results of his research accessible to people with primarily qualitative backgrounds. To that end, he will delve into the political and philosophical implications of his research as much as its mathematical implications. And, if time permits, he will also try to make a user-friendly library in Python with the tools he has made.
Philos is passionate about the intersection of philosophy and math, which is why he is excited to be studying algorithmic bias in artificial intelligence models at JURIST this summer. His previous experience has been in expository projects in abstract algebra and complex analysis, as well as research in astrophysics. In his free time, he enjoys exercising, playing the piano, and watching Netflix.