Elizabeth Mieczkowski

I am a PhD candidate advised by Tom Griffiths and Natalia Vélez in the AI & ML area of the Department of Computer Science at Princeton University. I am supported by the Department of Defense National Defense Science and Engineering Graduate Fellowship (NDSEG) and the Gordon Y. S. Wu Fellowship in Engineering.

I study multi-agent teams, whether human, AI, or both. Specifically, my goal is to develop a formal framework drawing on tools such as distributed computing and queueing theory to characterize the optimal strategies and fundamental trade-offs that emerge when multiple agents work together. I test these models in large-scale multi-player human experiments, LLM teams, and multi-agent reinforcement learning. Because this framework is mechanistic, it both generates testable predictions about teams and prescribes concrete improvements to performance. So far, it has shed light on problems such as when RL agents and people specialize during collaborative tasks, scalability, consistency, and architectural constraints on LLM teams, and when collaborators should contribute vs. remain idle.

Before starting my PhD, I spent two years as a lab tech with Nancy Kanwisher at MIT, where I studied similarities and divergences between CNNs and visual representations in the human brain. I received my B.A. in Computer Science with a minor in Psychology from Cornell University in 2021, where I conducted research in autonomous navigation and natural language processing. In 2019 and 2020, I was a software engineering intern at The New York Times.

Google Scholar // CV