Ari Holtzman

Ari is an incoming Assistant Professor of Computer Science and Data Science, starting July 2024.

His research has focused broadly on generative models of text: how we can use them and how can we understand them better. His research interests have spanned everything from dialogue, including winning the first Amazon Alexa Prize in 2017, to fundamental research on text generation, such as proposing Nucleus Sampling, a decoding algorithm used broadly in deployed systems such as the OpenAI API. With the new wave of powerful generative models being continually released, Ari has argued for using the lens of Complex Systems to understand generative models of human media, suggesting that a lack of precise behavioral vocabulary to describe what language models are doing is the bottleneck to explaining how language models are capable of such impressive performance on a range of tasks. He completed his PhD in Computer Science at the University of Washington studying “Interpretation Errors” in how we understand generative models after an interdisciplinary degree at NYU combining Computer Science and the Philosophy of Language.