Fighting Disasters with AI

We connected with PNNL, an impressive group.

DOE Announces ‘First Five Consortium’ to Fight Natural Disasters with AI
Oliver Peckham in Datanami

As wildland fires tear across California and hurricane season starts to warm up, natural disasters are top-of-mind for many Americans. Predicting and managing these disasters is an ongoing challenge, and researchers are leveraging technology from supercomputing to big data analytics to try to bridge these gaps. Now, the Department of Energy (DOE) has announced the First Five Consortium: a group of leaders in the AI space determined to use intelligent tools to combat natural disasters in the United States.

The consortium, co-chaired by the DOE and Microsoft, was formed in response to a January White House forum focused on disaster responses and is named after the “critical first five minutes in responding to a disaster.” The areas it hopes to tackle include wildfire prediction and fire line containment; damage assessment; search and rescue; and natural disasters like hurricanes and tornadoes.

In support of the consortium, Microsoft has established a “critical infrastructure team” that will use AI, confidential computing, advanced communications and more to improve disaster resilience.

“Artificial intelligence enables us to address some of humanity’s greatest challenges, and in this case, improve disaster resilience for populations around the world,” said Toni Townes-Whitley, president of U.S. regulated industries for Microsoft. “As evidenced by this consortium we have joined with the DOE and [the Department of Defense], it’s critical that private and public sectors work together to provide first responders with technology that has the potential to save lives.”

As an example project, the DOE highlighted ongoing work from Pacific Northwest National Laboratory (PNNL) and the Joint Artificial Intelligence Center (JAIC) in the Department of Defense (DoD). The researchers are scaling a prototype that uses deep learning algorithms to provide close-to-real-time data for first responders.  ... "