Large language models can write essays, hold conversations, and in general produce language that seems very human-like. But what lies behind this fluency? To what extent do their responses reflect linguistic abilities comparable to those of human speakers?

We use linguistic theory and methods from language research to examine the capabilities and limitations of AI. How do models respond to grammatical contrasts? Can they interpret meaning in context? And when their responses resemble human judgments, does that resemblance hold up across different examples and tasks?

These questions connect empirical testing with broader discussions about language, cognition, and what it means to understand. Through our research and interdisciplinary conversations at events such as our AI and Language Symposium, we explore what linguistics can reveal about AI and what AI might prompt us to reconsider about human language.