Katherine Van Koevering

Large language models are trained on human language — all of it. That means they absorb not just our vocabulary and grammar, but our biases, our behavioral quirks, our social patterns. I study what that means. I’m a postdoctoral researcher at the Johns Hopkins Data Science and AI Institute, working with Anjalie Field in the Field Lab. My research examines how sociolinguistic variation in prompts shapes LLM outputs, why AI systems exhibit distinctly human behavioral patterns, and how humans and AI systems adapt to each other over time. I received my PhD in 2025 from Cornell University, advised by Jon Kleinberg.

Research interests: AI linguistics  ·  LLM behavior  ·  human/AI communication  ·  human-AI co-adaptation  ·  language variation  ·  computational linguistics  ·  network science