Using AI to Understand Public Response in a Crisis
During a crisis, clear communication can make a meaningful difference. But with thousands,
or even millions, of people responding online, understanding what the public is asking,
sharing and reacting to can be difficult.
Understanding Crisis Communication at Scale
Prachi Aswani, an undergraduate researcher at the University of Utah, is exploring
how AI can help.
Aswani is part of the team behind CrisisIntel, a research project using AI and natural
language processing to study public responses to official crisis communication on
social media.
The research focuses on an important distinction: discussion versus reaction. CrisisIntel
examines whether responses contribute questions, information or meaningful dialogue,
or reflect more immediate reactions. The framework also helps researchers study themes,
misinformation signals and communication patterns across large volumes of online conversation.
Looking Beyond Traditional Metrics
Using more than 1.27 million COVID-19 social media posts and replies, Aswani and her
collaborators are studying patterns that can be difficult to see through traditional
measures such as likes, replies or sentiment alone.
Research with Real-World Impact
The work could ultimately help public-health agencies, emergency managers and other
communicators better understand what people are asking, where confusion or misinformation
may be growing and what information communities need during a crisis.
The approach also has potential beyond public health, including wildfires, natural
disasters and other emergencies where timely, trustworthy communication matters.
Aswani is conducting the research alongside Marina Kogan of the Kahlert School of Computing, Di Wang and Mike Kirby of the Scientific Computing and Imaging Institute and One-U Responsible AI Initiative. She has also shared the work through a poster presentation at Research on Capitol
Hill, bringing the research and its potential public impact to a broader audience.
It is an impressive example of how AI can be applied to a complex, real-world challenge,
with the potential to help communities better navigate moments when clear information
matters most.
For more information or to express interest in collaborating, email Prachi.