How Government Might Use AI to Improve Process, Navigate Disaster

In a recent discussion of using artificial intelligence to improve the state's disaster resilience, Bijan Karimi of San Francisco’s Department of Emergency Management examined its value and where public-sector needs could exist.

This story is limited to Techwire Insider members.
This story is limited to Techwire Insider members. Login below to read this story or learn about membership.
State and local governments are not unfamiliar with artificial intelligence, but in recent remarks, a consolidated city-county emergency leader offered examples of ways governments might use it to fine-tune their disaster response — and, in turn, how the private sector might work with them.

Bijan Karimi, assistant deputy director at the Division of Emergency Services in the San Francisco Department of Emergency Management, provided examples of how technologies — even some that aren’t bleeding edge — can enhance public-sector response and potentially circle in the private sector. Among the takeaways, Karimi discussed how AI can:

• Help staffers extend their reach. The public sector faces a continual onslaught of valuable data, Karimi said Thursday during a discussion of “Into the Storm: Using Artificial Intelligence to Improve California’s Disaster Resilience,” a new report from Microsoft and the Partnership for Public Service. His team’s job at San Francisco’s Emergency Operations Center, he said, is “to gather, analyze and disseminate information,” adding: “And I can’t get my folks enough time to focus on the, ‘So what does it mean?’ Or in the abstract, reasoning. And that’s why hopefully we can use AI to help us in that capacity more.”

• Help governments fine-tune resource deployment. Karimi cited an example of people using AI to compare before-and-after photos following storm damage and said AI that can do A/B comparisons, and is also trained to identify what disasters and damages look like with real-time satellite imagery, can be very valuable, “so that I can put shelters where they’re needed, or I know where infrastructure is more damaged and I can get resources there more quickly.”

• Minimize effort duplication. His organization has migrated to Microsoft Teams, Karimi said, but individual chats still take place via other methods of communication. Another “AI platform that can sit on top of Teams” and notify leaders when two discussions on the same topic are occurring would be helpful, he said.

“The more that I can see what’s going on in my organization by an agent that’s working in the background, looking for common themes, common queries, that’s going to help me.”

• Vendors might want to consider a particular government’s privacy concerns before initiating a conversation about solutions. In San Francisco, the assistant deputy director said, privacy is a “huge concern,” and rules in place inform data and camera usage, and limit active surveillance. The ability to wield technology during an incident could help the public sector do its job better, Karimi said, but its use of that tech could also be subject to “a certain set of constraints.”

• Not everyone has a chatbot, even now — although state agencies using them include Covered California, the California Department of Motor Vehicles, the California Secretary of State’s Office and the Financial Information System for California (FI$Cal). The 2018 Camp Fire demonstrated the technology’s value to San Francisco, Karimi said, as it saw North Bay neighbor Sonoma County grapple with high call volumes during the disaster.

“They know they’re getting a computer response but that provides some amount of connection,” he said of residents using chatbots.

Theo Douglas is Assistant Managing Editor of Techwire.