John Thickstun

John Thickstun is an Assistant Professor of Computer Science at Cornell University, effective Fall 2024. Previously, he was a Postdoctoral Scholar at Stanford University under Percy Liang, completing his PhD at the University of Washington's Allen School of Computer Science & Engineering under Sham Kakade and Zaid Harchaoui. He holds a BA in Applied Mathematics from Brown University. His research focuses on machine learning, generative models, and their societal implications, with notable work on watermarking language models (TMLR 2024), music generation via the Anticipatory Music Transformer (TMLR 2024), and evaluation metrics like MAUVE (JMLR 2023). He has received awards including the ACL Outstanding Paper (2023) and NeurIPS Outstanding Paper (2021). His work bridges technical innovation with applications in music technology and AI policy.

Education:

• PhD in Computer Science, University of Washington (2023)
• BS in Applied Mathematics, Brown University (2017)

Research Interests:

• Controllable generative models (text, music)
• AI ethics and provenance via watermarks
• Music generation and analysis (MusicNet dataset)
• Evaluation metrics for generative systems

Notable Publications:

• "Robust Distortion-free Watermarks for Language Models" (TMLR 2024)
• "Anticipatory Music Transformer" (TMLR 2024)
• "MAUVE Scores for Generative Models" (JMLR 2023)

Awards:

• ACL Outstanding Paper (2023)
• NeurIPS Outstanding Paper (2021)