[CS] Kevin Bryson Candidacy Exam/Dec 10, 2025

[via cs]

This is an announcement of Kevin Bryson's Candidacy Exam.
===============================================
Candidate: Kevin Bryson

Date: Wednesday, December 10, 2025

Time:  1 pm CST

Remote Location: https://uchicago.zoom.us/j/98080835076?pwd=2JjU9Gdia6RW4gQ6xtjrmihQswtcf3.1&jst=2

Location: JCL 223

Title:  Byproducts of Complex Algorithmic Systems: Fairness, Usability, and Transparency Challenges

Abstract: Algorithmic systems are pervasive in daily life, automatically making decisions about content on social media feeds as well as mortgage approvals. These automated systems are highly desired because of their efficiency and efficacy gains over manual efforts and in some cases, their purported objectivity. Deployments of algorithmic systems have repeatedly faced scrutiny from the broader public and academics when found to be discriminatory, creepy, or illegal. In spite of efforts to address these issues, the repetition of these events suggests that the production and deployment of algorithmic systems necessitates certain byproducts. Namely, usability, fairness, and transparency challenges.

This thesis explores the effects of these byproducts on three groups of people. First, we explore user interactions with personalized advertising on large online platforms, in particular, we conduct a user study exploring the usability and usefulness of the ad transparency systems---the settings and information provided about personalized advertising. Second, we turn to data scientists and explore their role in achieving fair outcomes when developing machine learning models, but particularly during the data preparation phase. In interviews with data scientists, we focus on an under-explored grouping of techniques which modify the training data to ask an underlying provocative question, if the training data is discriminatory, why not modify it before learning from it? Finally, we audit TikTok's personalization algorithm by investigating how user's help-seeking behaviors on TikTok impact and are impacted by algorithmic personalization through a data donation study. 

Advisor: Blase Ur

Committee Members: Blase Ur, Alex Kale, Damon McCoy