Better use of cloud can improve outlook for agencies’ AI development, White House report says

The Select Committee on Artificial Intelligence has recommendations for agencies to better utilize cloud computing in federally funded AI research and development.
cloud computing, AI
(Getty Images)

Federal agencies may not be using cloud computing effectively as they try to develop artificial intelligence capabilities, according to a report released Tuesday from the White House’s advisory panel on artificial intelligence.

“While it is becoming well recognized that cloud computing can advance AI R&D, several technical and administrative challenges are currently limiting cloud adoption, ” the report from the Select Committee on Artificial Intelligence says.

For example, inconsistencies in the methods of accessing and using cloud computing create barriers to using the technology for AI research, the report says. It also cites limited access to education and training opportunities for cloud use.

The committee, part of the National Science and Technology Council (NSTC) at the Office of Science and Technology Policy, made four key recommendations to overcome these challenges. They revolve around pilot projects, education, identity management and interoperability among different cloud platforms.


The panel also suggested that pilot projects can help explore the challenges and advantages of using cloud computing in federally funded AI research, something that agencies like the Department of Defense’s Joint Artificial Intelligence Center, the Department of Transportation, and the Department of Veterans Affairs have already launched.

Agencies should then couple these projects with education and training opportunities for researchers to help “realize [cloud-based platforms’] full benefits for advancing AI frontiers,” the panel said.

Also in that timeline, the committee suggested that the NSTC’s Subcommittee on Machine Learning & Artificial Intelligence establish a task force to find best practices in identity management, such as token-based access, and single-sign-on strategies. The goal would be to respect data contributors while also allowing for discovery and research within relevant datasets.

“It is time to move beyond role-based access to data to new strategies that promote investigator access to data in a manner that provides fine-grained access control,” the report reads.

The select committee was formed in May 2018 with the mission to ensure U.S. leadership in artificial intelligence.

Latest Podcasts