The Department of Energy and the Department of Commerce will collaborate in research and development projects on safe artificial intelligence use under a memorandum of understanding signed recently. 

The Commerce Department will be represented in the partnership by the National Institute of Standards and Technology, home of the U.S. AI Safety Institute, tasked under the National Security Memorandum on AI as a key government hub for safety efforts on the technology, DOE said Wednesday.

Besides facilitating joint research and information sharing, the MOU will also enable the DOE and its national laboratories to provide technical support to the US AISI and NIST, the department added.

Focus Areas and Leaders of the Collaboration

The activities that will be pursued under the DOE-DOC agreement include evaluating AI models’ impact on public safety, such as risks to energy security and critical infrastructure. The key tasks for the risk analyses extend to such efforts as developing and evaluating privacy-enhancing solutions to protect personal and commercial proprietary information.  

The DOE and DOC appointed their respective principal coordinators to facilitate the departments’ collaboration on AI safety. The appointees are Helena Fu, DOE director of the newly created Office of Critical and Emerging Technologies, and Elizabeth Kelly, US AISI director as the Commerce Department coordinator.

The Department of Veterans Affairs and the Food and Drug Administration have agreed to jointly develop a virtual laboratory where promising healthcare tools powered by artificial intelligence can be tested and evaluated, Nextgov/FCW reported Wednesday.

Speaking at the Veterans Health Administration Innovation Experience conference, VA Undersecretary for Health Shereef Elnahal announced the partnership, noting that the interagency testing lab will provide selected parties a safe and secure location to prove their technologies in accordance with the October 2023 executive order promoting the responsible adoption and use of new AI capabilities.

Elnahal said the virtual environment will ensure that AI technologies are safe and effective for veterans and patients and adhere to trustworthy AI principles.

The VA and the FDA are now preparing to establish the virtual proving ground, which is expected to be completed in six months.

According to FDA Commissioner Robert Califf, the testing lab will be the first intergovernmental health AI laboratory, adding that it will support the development of approaches for assessing safety and performance metrics of AI-enabled healthcare products. 

Engage with industry luminaries, thought leaders and experts at the 2024 Healthcare Summit on Dec. 11 as they delve into critical issues such as the rise of artificial intelligence.

SandboxAQ will partner with Sanofi, a global pharmaceutical and healthcare company, to locate biomarkers during clinical development.

Using SandboxAQ’s quantitative AI models, the project aims to filter knowledge graphs to gain a better understanding of human biology, supporting scientists’ ability to promote trust throughout investigational medicines and clinical development, the AI and quantum technology company announced Tuesday.

Nadia Harhen, general manager of AI simulation at SandboxAQ, said, “The application with Sanofi focused on biomarker identification is very exciting as it increases our reach into later-stage clinical development and the patient benefit that can be unlocked far beyond the early stages of drug discovery.”

SandboxAQ’s Large Quantitative AI Models 

SandboxAQ’s large quantitative models — or LQMs — are trained on numerous data streams including proprietary data produced by the company’s algorithms. SandboxAQ’s LQMs avoid the barriers in scale and accuracy critical to natural large language models trained on public data accessible on the internet.

“Large Quantitative Models, such as those we use to causally filter through knowledge graphs, are proving impactful in many areas across the life sciences, from drug repurposing to reverse screening,” Harhen noted.

U.S. Army Contracting Command – Aberdeen Proving Ground is conducting a pilot test of A.I. Flow, a tailored AI workspace built by the Communications-Electronics Command Software Engineering Center, or CECOM SEC, to enhance the Army’s operations.

A.I. Flow, authorized to operate with controlled unclassified information, is currently accessible to select engineering center personnel, the U.S. Army said Tuesday.

A.I. Flow Uses

The product, which has a secure, user-friendly interface, can be used for various applications, including writing appraisals, new job information and contracts. CECOM SEC, in partnership with ACC-APG, sets parameters, uploads example documents and creates guidelines to create a justification and approval document, or J&A, for military contracting. According to the military branch, the trial showed that A.I. Flow can deliver one such document that is 80 percent complete.

With the AI-powered system, developing a J&A document will be faster compared to the manual process, which takes days or weeks.

A.I. Flow can be customized to meet the specific needs of users. The team behind the software’s development continuously receives feedback to improve the program and support more use cases.

Generative AI as a Service

The pilot test is expected to result in a fully operational A.I. Flow, which Ronald Rizzo, deputy executive director of CECOM SEC, described as “generative AI as a service.”