The CDS Monthly Research Feature |
Paying employees a fixed bonus to work exclusively for one employer can backfire, triggering a “pure income effect” where higher wealth leads to reduced working hours. Incoming CDS Assistant Professor of Economics and Data Science Jaume Vives-i-Bastida explores this economic phenomenon in a new paper that analyzes the labor supply of physicians in Spain. The study, titled “Pushing Back Against Private Practice: The Unintended Effects of Paying Public Doctors More,” was co-authored with Jonathan Gruber of MIT, Núria Mas of IESE Business School, and Judit Vall Castelló of the Universitat de Barcelona.
|
|
|
Artificial intelligence systems, much like humans, struggle to objectively evaluate their own work when they rely on the same internal logic to both solve a problem and verify the answer. This inherent “self-enhancement bias” suggests that a single dominant AI model may hit a ceiling in self-improvement, necessitating an ecosystem of diverse models to keep each other honest. In a new paper, “When Does Verification Pay Off? A Closer Look at LLMs as Solution Verifiers,” Courant PhD student Jack Lu, CDS PhD student Ryan Teehan, CDS MS student Jinran Jin, and CDS Assistant Professor of Computer Science and Data Science Mengye Ren provided empirical evidence for this.
|
The mathematical landscape governing the optimal settings of a deep learning model is not a chaotic mess, but looks much like a simple, orderly quadratic equation. Courant PhD student Nick Lourie used this surprising insight to derive a new theorem that removes the guesswork from training artificial intelligence, offering researchers a statistical “microscope” to analyze their experiments with unprecedented clarity. Lourie, along with his advisors CDS Professor of Computer Science and Data Science Kyunghyun Cho and CDS Associate Professor of Computer Science & Data Science He He, detailed this finding in their paper, “Hyperparameter Loss Surfaces Are Simple Near Their Optima.”
|
For decades, the United States held an unshakeable lead in the technologies that map our world — the sensors and satellites that underpin everything from autonomous vehicles to national defense — but that lead has not just vanished; it has been completely inverted. A new study by NYU Tandon Professor of Civil and Urban Engineering Debra Laefer and CDS MS alum Jingru (Catherine) Hua quantifies a massive geopolitical shift in the field of remote sensing, revealing that while the U.S. produced 88% of the field’s research papers in the 1960s, it now contributes less than 9%, while China’s share has skyrocketed to nearly half.
|
|
|
Copyright (C) 2026 Center for Data Science. All rights reserved.
|
|
|
|
