Skip to main content
ORNL’s Suhas Sreehari explains the algebraic and topological foundations of representation systems, used in generative AI technology such as large language models. Credit: Lena Shoemaker/ORNL, U.S. Dept. of Energy

In the age of easy access to generative AI software, user can take steps to stay safe. Suhas Sreehari, an applied mathematician, identifies misconceptions of generative AI that could lead to unintentionally bad outcomes for a user. 
 

Peter Thornton, second from right, is director of ORNL’s Climate Change Science Institute. He shares insights on the regional impacts of changing weather patterns during the Second Annual Appalachian Carbon Forum in Lexington, Kentucky. Credit: University of Kentucky’s Center for Applied Energy Research.

ORNL hosted the second annual Appalachian Carbon Forum in Lexington March 7-8, 2024, where ORNL and University of Kentucky’s Center for Applied Energy Research scientists led discussions with representatives from

Intern Noah Miller, left, and his mentor, Joe McVeigh, stand with their poster at the American Glovebox Society conference in 2023.

College intern Noah Miller is on his 3rd consecutive internship at ORNL, currently working on developing an automated pellet inspection system for Oak Ridge National Laboratory’s Plutonium-238 Supply Program. Along with his success at ORNL, Miller is also focusing on becoming a mentor for kids, giving back to the place where he discovered his passion and developed his skills. 

Sean Oesch

While government regulations are slowly coming, a group of cybersecurity professionals are sharing best practices to protect large language models powering these tools. Sean Oesch, a leader in emerging cyber technologies, recently contributed to the OWASP AI Security and Privacy Guide to inform global AI security standards and regulations.

AI-driven attention mechanisms aid in streamlining cancer pathology reporting.

In partnership with the National Cancer Institute, researchers from the Department of Energy’s Oak Ridge National Laboratory’s Modeling Outcomes for Surveillance using Scalable Artificial Intelligence are building on their groundbreaking work to

Instantaneous solution quantities shown for a static Mach 1.4 solution on a mesh consisting of 33 billion elements using 33,880 GPUs, or 90% of Frontier.  From left to right, contours show the mass fractions of the hydroxyl radical and H2O, the temperature in Kelvin, and the local Mach number. Credit: Gabriel Nastac/NASA

Since 2019, a team of NASA scientists and their partners have been using NASA’s FUN3D software on supercomputers located at the Department of Energy’s Oak Ridge Leadership Computing Facility to conduct computational fluid dynamics simulations of a human-scale Mars lander. The team’s ongoing research project is a first step in determining how to safely land a vehicle with humans onboard onto the surface of Mars.

New system combines human, artificial intelligence to improve experimentation

To capitalize on AI and researcher strengths, scientists developed a human-AI collaboration recommender system for improved experimentation performance. 

A multidirectorate group from ORNL attended AGU23 and came away inspired for the year ahead in geospatial, earth and climate science

ORNL scientists and researchers attended the annual American Geophysical Union meeting and came away inspired for the year ahead in geospatial, earth and climate science. 

: ORNL climate modeling expertise contributed to an AI-backed model that assesses global emissions of ammonia from croplands now and in a warmer future, while identifying mitigation strategies. This map highlights croplands around the world. Credit: U.S. Geological Survey

ORNL climate modeling expertise contributed to a project that assessed global emissions of ammonia from croplands now and in a warmer future, while also identifying solutions tuned to local growing conditions.

Conversion of an atomic structure into a graph, where atoms are treating as nodes and interatomic bonds as edges. Credit: Massimiliano “Max” Lupo Pasini/ORNL, U.S. Dept. of Energy

Researchers at the Department of Energy’s Oak Ridge and Lawrence Berkeley National Laboratories are evolving graph neural networks to scale on the nation’s most powerful computational resources, a necessary step in tackling today’s data-centric