How AI boosts emergency response in the new age of super disasters
Artificial intelligence can help emergency managers save lives, limit economic impact and provide essential disaster relief
As one-in-1000 year weather events strike with stomach-churning regularity, the new normal is not normal.
Hurricanes Harvey, Irma and Maria devastated cities and countries in a single one-month period. That same month, two earthquakes, including their strongest in a century, hit Mexico; and in the worst wildfire season in memory, “ash fell like snow” in the Western United States.[1]
The super disasters have strained emergency services beyond capacity, and required the heroic help of both governments and citizens. As we look to assist those in need and continue the recovery process, it’s important to look at how powerful new technology can better respond to these emergencies in the future.
How artificial intelligence augments overwhelmed dispatch centers
During major disasters like Hurricane Harvey, dispatch centers are inundated with calls. Houston Mayor Sylvester Turner said 911 operators received 56,000 calls in less than 24 hours during one of the first days of the disaster.[2]
This mammoth bottleneck prevents dispatchers from quickly directing first responders to people in need who don’t have the time or phone battery life to keep calling for help.
Artificial intelligence can serve as a 24/7 dispatcher. No need for gulps of bitter coffee to power through stressful emergency situations. AI can process and accelerate calls to dispatchers while filtering out redundant or less urgent calls. It can interact with callers naturally, instantly transcribe and translate languages and analyze the tone of voice for urgency.
“AI provides call triage,” said Bill Josko, IBM’s GBS Public Safety Practice Leader for the U.S. “When 911 centers are overrun, AI can analyze massive amounts of calls, determine what type of incident occurred and verify the location. It listens to the content of calls to prioritize the emergency, such as a fender bender versus senior citizens stuck in a house with five feet of water.”
During Orlando’s Pulse nightclub massacre, victims had to stay quiet to not draw attention. That meant they couldn’t call for help, but also couldn’t text 911, because most emergency dispatch centers in the US aren’t equipped to receive text messages, photos and videos — or tap into the detailed location services of mobile phones.[3] Instead, they texted family and friends to call 911.
Many governments globally, however, are currently upgrading their emergency dispatch technology to receive more types of data, such as the Next Generation 911 initiative in the US and Canada. Dispatch centers augmented by AI can ingest the data from not just calls but text, video, audio, pictures and millions of social media posts to make quick assessments.
The insights gleaned from all this information can be fed to emergency response teams out in the field. Details about a family, their ages, medical needs and precise location can automatically populate the connected devices of first responders.
Mass changes in user sentiment on social channels could indicate an emergency is occurring even before an incident — like a train derailment or tornado — is reported, giving first responders a critical head start.
No fake news — making social media storm proof
Hurricane Harvey was considered the US’s first “social media storm”–where social media helped the government as well as private citizens rescue those in need. Nursing home residents were rescued after a photo of them stuck in waist-high floodwater went viral.
First responders don’t, however, have the bandwidth to monitor all social posts. Social media is not just a fire hose of information–key details can be missing or inaccurate. During emergency events, Twitter can both reflect the humming community spirit and resemble the frenzied just-swatted antagonism of a beehive.
“Using social media as a platform for emergency calls can lead to calls for help getting overlooked, misidentification, the spread of false information and even fraud,” according to TechCrunch.[4] What’s needed is a super journalist/police investigator hybrid that never sleeps.
“Bad guys thrive in chaos. And news is not the gospel truth just because somebody tweeted it,” said Josko. “AI validates which information is real and which is fake. It sniffs out the ‘ground truth’ by looking at all the other sources contradicting what people are saying in social media.”
If someone tweets an image that shows a flooded airport, AI can instantly confirm the geo-location and grab sensor data, photos, and surveillance or drone videos to quickly verify if that image is real or fake, such as an image taken during a different, unrelated disaster.
AI can also view the unstructured data in pictures and videos posted to social channels to find missing people. It can search millions of hours of video footage for a person with blonde hair, a red shirt, and a beard, and then cross-check that against surveillance, drone or body-worn camera video, pinpointing where the missing person was last located and at what time. The next time they are spotted, an alert is triggered so emergency managers can confirm their safety — or deploy a rescue operation.
Learn more about IBM’s emergency management & public safety solutions and see how technology can help with super disasters.
[1] https://www.seattletimes.com/seattle-news/ash-falls-like-snow-in-seattle-as-wildfires-rage/
[2] http://www.latimes.com/nation/la-na-tropical-storm-harvey-20170827-story.html
[3] http://www.pewtrusts.org/en/research-and-analysis/blogs/stateline/2016/07/29/cities-states-plod-toward-next-generation-911
