am Carey Okal, a master's student in Information Science at Cornell University (B.S. May 2026). My research asks when large language models can stand in for people in social science research. I study this through LLM-agent simulations of social behavior and LLM personas as survey respondents, and I have also built large-scale datasets for studying persuasion in political news and worked on data science for humanitarian response. This website is a window into my work.
At the Social Dynamics Lab, I study how social order emerges when LLM agents are cast as guards and prisoners in a simulation of the Stanford Prison Experiment. I developed the coding framework the study uses to analyze guard behavior, and I am now testing whether telling agents that they and their counterparts are human or AI changes how they exercise authority.
In a co-authored working paper, we test whether LLM personas built from real General Social Survey respondents react to survey design the way people do, using a randomized experiment that showed some respondents a "No opinion" option and hid it from others as the human benchmark.
I built the roughly 3-million-article corpus of U.S. news articles (14 national and local outlets, 1996–2025) the lab uses to study how persuasive strategies vary with outlet credibility and ideology, collecting article links from GDELT and the 3DLNews dataset, matching them to full text, and standardizing them into a unified schema.
The Data Science to Build Resilience and Improve Humanitarian Response Thought Summit, held at Cornell in May 2026, convened researchers and humanitarian leaders from institutions including the World Bank, the World Food Programme, UNHCR, IFPRI, NASA JPL, and Kenya's National Drought Management Authority.
Community conservancies in East Africa compete for visibility against booking platforms that take large commissions. Through Cornell's SMART Fellowship, I built a direct-booking and marketplace platform for the Il Ngwesi conservancy, which won our cohort hackathon and was adopted by the Kenya Wildlife Conservancies Association as a blueprint for expansion to 24 Maasai community conservancies.
I've taught at Cornell since Fall 2023, supporting around 300 students per semester across machine learning, data science, and web development courses, and earned perfect 10/10 TA evaluations all semesters.
Graduate Teaching Assistant · INFO 4555/5556 · FA 26
I'm joining the course staff for INFO 4555/5556, an upper-level business intelligence course. I'll support students through the work sessions and the semester-long team project that anchors the course, holding office hours and grading their work as they move from dimensional modeling and data transformation into metadata and data governance.
Course Assistant · ORIE 3120 · SP 26
I led weekly recitation sections for ORIE 3120, a large applied data science course in Cornell Engineering, covering everything from SQL and Tableau through regression, forecasting, and A/B test design. Outside of section, I held office hours, graded homework and project milestones, and kept an eye on Ed Discussion so students got help between classes.
Teaching Assistant · INFO 2950 · FA 24
I taught weekly lab sections for INFO 2950, Cornell's introductory data science course. Most of my time in lab went toward Python, statistical analysis, and helping students build the confidence to debug their own work when something broke. I held office hours, responded to Ed Discussion posts within a day, and met weekly with the professor to relay student feedback and refine the course.
Teaching Assistant · INFO 1300 · FA 23, SP 24, FA 25
I taught weekly lab sections for INFO 1300, an introductory web design and programming course. In lab, students designed and shipped their first accessible websites, working from markup and layout through basic UX decisions. I also held office hours, graded exams and projects, and answered questions on Ed Discussion throughout the week.