NYU faculty selected for three DOE Genesis Mission AI research projects

The projects were among 278 chosen from more than 5,000 applications and will explore quantum computer control, scientific hypothesis generation and decentralized AI agents

Three projects involving NYU faculty will investigate AI applications in quantum computing and scientific discovery

New York University faculty have joined three US Department of Energy Genesis Mission projects developing AI systems for scientific research, including tools for programming quantum computers and generating evidence-based hypotheses.

The three projects bring together researchers from NYU Arts & Science, the Courant Institute School of Mathematics, Computing, and Data Science, and the Tandon School of Engineering. Their collaborators include Brookhaven National Laboratory, Argonne National Laboratory, Carnegie Mellon University, Iowa State University and Microsoft.

More than 5,000 applications were submitted to the first round of the Genesis Mission, with 278 projects selected. The national DOE initiative connects national laboratories, universities and industry partners with AI, high-performance computing, scientific instruments and data infrastructure.

Juan J. de Pablo, Executive Vice President for Global Science and Technology at NYU and Executive Dean of the Tandon School of Engineering, highlighted the range of work in a LinkedIn post. “Their selection reflects the extraordinary caliber of research taking place across NYU and the power of bringing together expertise from multiple disciplines to tackle the challenges of the future,” he wrote.

AI to control quantum hardware directly

One of the three projects will test an alternative approach to programming quantum computers.

“AI-driven Pulse-level Optimal Control of Many-body Quantum Eigensolvers” is led by Andrei Vrajitoarea, Assistant Professor of Physics, and Norah Hoffman, Assistant Professor of Chemistry and Physics.

Quantum algorithms are typically written as abstract instructions before being translated into a fixed sequence of hardware operations. According to NYU, that translation can slow computation and introduce errors.

Vrajitoarea and Hoffman’s team will instead use AI to control the quantum hardware directly, removing that intermediate translation stage.

De Pablo described the project more simply on LinkedIn, writing that the two researchers are “rethinking how we program quantum computers, using AI to control the hardware directly.”

MARS will build teams of hypothesis-generating AI agents

A second project, MARS, will examine how specialized groups of AI agents could support scientific hypothesis generation.

The project’s full name is Multi-Agent Reinforcement Learning for Scientific Hypothesis Generation. Brookhaven National Laboratory is leading the work in collaboration with Carnegie Mellon University, NYU and Microsoft.

Mengye Ren, Assistant Professor of Computer Science and Data Science at NYU’s Courant Institute School, will help develop teams of agents designed to generate evidence-based hypotheses more efficiently.

Initial work will focus on two areas: identifying problems in analog circuit designs and interpreting material structures from X-ray data.

While the project is intended to support stronger scientific hypotheses, no evidence of its performance has yet been supplied. The announcement outlines the planned research rather than presenting completed findings.

DAISY agents will divide work and exchange knowledge

The third project shifts the emphasis from teams managed around a specific hypothesis task to a decentralized network of AI agents.

DAISY, the Decentralized Agentic Intelligence System for Scientific Inquiry, is led by Iowa State University in collaboration with Argonne National Laboratory.

NYU participants include Chinmay Hegde, Associate Professor of Computer Science and Engineering, and Juliana Freire, Institute Professor of Computer Science and Engineering at NYU Tandon School of Engineering.

The team plans to develop agents that divide work and share compact knowledge tokens. NYU will help lead the creation of the platform’s core toolkit.

Early applications are planned in materials science, crop breeding for bioenergy and the power grid.

De Pablo positions the three projects as part of a change in how research is conducted. “Science is entering an era where discovery itself can be accelerated by the tools we build to pursue it,” he says. “This is exactly the kind of cross-disciplinary momentum the Genesis Mission was designed to unlock, and it’s where NYU is ready to lead.”

Gerard Ben Arous, Dean of NYU’s Courant Institute School, identifies the development of AI agents into scientific applications as a continuing institutional priority.

“Leveraging AI from the ground up is vital for both scientific discovery and technological advancement,” he explains. “Developing promising AI agents into advanced applications has been a driver of the Courant Institute School’s success, and it will remain a core focus under the Genesis Mission.”

The three NYU projects are part of the inaugural group of 278 Genesis Mission selections. No research schedule, funding allocation or project results were disclosed.