The CDS Monthly Research Feature |
A robot can practice in one environment for millions of steps and still encounter something new when the lighting changes, an object weighs less than expected, or a familiar task takes place in a different space. CDS PhD student Ying Wang, working with NYU postdoctoral researcher Oumayma Bounou, CDS founding director Yann LeCun, and CDS Assistant Professor Mengye Ren, developed a way for a world model to respond to those mismatches while it was planning, rather than relying entirely on what it learned before deployment.
|
|
|
Tracking a patient’s movements during stroke rehabilitation sounds simple, but doing it accurately and automatically remains difficult, and even sophisticated AI models can struggle to describe a hand’s movements and object interactions. In a new study, CDS PhD student Victor Li, CDS Associate Professor of Mathematics and Data Science Carlos Fernandez-Granda, and their collaborators tested whether vision-language models could extract useful clinical information directly from videos of people undergoing stroke rehabilitation.
|
In 2025, Anthropic reported that its Claude models could detect when researchers injected a concept directly into their neural activity. Anthropic presented this result as evidence of “emergent introspective awareness.” However, earlier this year, CDS PhD student Shashwat Singh, CDS Associate Professor of Linguistics and Data Science Tal Linzen, and CDS Faculty Fellow Shauli Ravfogel challenged that conclusion.
|
We are currently hiring a Faculty/Chair and Faculty Fellows for Fall 2027. The Chair will guide the growth of CDS, lead faculty recruitment, and pursue the Center's research vision. Faculty Fellows will lead research projects and work with faculty across CDS and the wider university.
|
|
|
Copyright (C) 2026 Center for Data Science. All rights reserved.
|
|
|
|
