Welcome to Antoni Czolgowski's homepage!

AI agent decision-making and agentic simulations

I’m a master’s student in data science at CU Boulder, with a background in quantitative economics from SGH Warsaw and experience building research software and AI systems.

I started working with language models at Orange just after GPT-3.5 was introduced. I remember how frustrating it was to get confidently wrong answers in Polish. That curiosity about where models go wrong eventually led me to research comparing their responses with human survey data, examining how well they represent different groups and how we can change it.

For my PhD, I want to move from studying what models say to understanding how agents decide and act. I'm interested in how they use information, adapt to changing conditions and coordinate with people and other agents. I want to build simulations where we can change something, follow the consequences and understand why the outcome changed. Organizations, cities and public policy give us useful settings for asking those questions and testing whether simulations can reliably inform real decisions.

I’m currently working on Agent reliability under disruption, studying whether agents representing real people can adapt when their daily routines change.

• Research: agent decisions, simulations and human–AI collaboration.
• Publications and talks: empirical studies of how language models represent human populations.
• Cultural Misalignment in Large Language Models: Detection, Measurement, and Mitigation Through Targeted Fine-Tuning. With Abel Iyasele. arXiv preprint, September 2026. PDF · Code and data.
• Projects: electoral behavior, simulated audiences and digital access.
• Software: research code, tools for AI agents and data infrastructure.
• Experience and teaching.

Contact: antoni.czolgowski@gmail.com
Google Scholar · GitHub · LinkedIn · Strava