Why AI must include community voices

[...]

First, tools that promise to streamline assessments and planning will miss critical information the community can provide if those tools work from existing datasets containing what is already known and documented. Second, planning is not just an output, but rather a mechanism for building relationships, galvanizing action, and spurring change within the community. Overreliance on AI can lead to missed opportunities. Finally, humans are best positioned to think in highly creative, out-of-the-box ways while ensuring that thinking about disaster issues is deeply rooted in a community context. AI is not a substitute for what humans can offer in this regard.

None of this is meant to suggest AI should be avoided in emergency management. The potential efficiency and effectiveness gains are real and could benefit the nation’s resilience. AI tools could reduce burdens and improve efficiencies in ways that maintain whole community engagement. Better yet, leaders should seek tools that can enhance such engagement in many ways:

• Assessment: AI can identify and synthesize large bodies of research or other data, which could then be shared with communities.
• Planning: AI could be used to support analysis of community engagement sessions, providing rapid feedback and identifying concerns.
• Management: AI can also reduce the onerous components associated with implementing emergency management plans. Once a community outlines its expectations related to hazard mitigation, for instance, AI can help develop mitigation grants.
• Quality Assurance: AI might be deployed to ensure that whole community representation is complete as mitigation and response plans are being developed.

Communities themselves are already using AI in these ways and can be a source of ideas. For instance, the authors have been engaged with one community group working on recovery from the 2025 Los Angeles County firestorms that is using AI to comb through the thousands of survivor messages in their online forum. AI can identify questions that require answers and ascertain needs that must be met. Another group is using AI for near-real-time analysis of fire recovery planning sessions and workshops they are hosting. Finally, in Northern California, the authors have also engaged with a network of community organizations that is leveraging AI to refine their wildfire resilience strategies—iterating with large language models as they home in on and articulate objectives and activities.

[...]