On August 5, Captain Kris Yeary was awakened around 3 a.m. to a wildfire in southern California. But unlike usual, the firefighters were not notified by an emergency call: the alert came from artificial intelligence. (Photo: Getty Images)
San Diego — On August 5, Captain Kris Yeary was awakened around 3 a.m. to a wildfire in southern California. But unlike usual, the firefighters were not notified by an emergency call: the alert came from artificial intelligence.
The officer immediately returned to the San Diego command center. On the slew of screens dedicated to monitoring the region, a column of smoke emanated from Mount Laguna, about 70 kilometers away.
The firefighter quickly sent personnel to put out the flames within an hour. As a result, the fire devoured barely a thousand square meters and never threatened the 1,600 residents of Pine Valley, the small town below.
“If the AI had not alerted us, it could have grown significantly,” the officer told AFP. “It could have been a devastating fire.”
Cradle of Silicon Valley, California is a pioneer in artificial intelligence, this technology which is based on algorithms capable of imitating certain human behaviors and improving on their own.
She is now using it to fight wildfires, which have killed more than 200 people in the past decade and are getting worse with climate change.
Since the end of June, software has continuously scanned 1,040 cameras, scattered at high points throughout the “Golden State” by the University of San Diego, and alerts firefighters from the Cal Fire fire agency when he thinks he spots smoke.
Promising results
The initial results are so promising that the system now equips every emergency command center in the state.
“We anticipate 911 calls in about 40% of cases. And it will get better,” explains Neal Driscoll, the researcher at the head of the ALERTCalifornia platform.
“Our indicator of success is the fires you never hear about,” adds this professor of geology and geophysics at the University of San Diego.
Concretely, each firefighter on duty responsible for scanning dozens of cameras in his area is now assisted by AI.
When the software believes it detects smoke, it displays a small red rectangle on the screen, with a percentage indicating its degree of certainty. It is then up to the operator to confirm the seriousness of the alert.
Because for the moment, the robot seems quite paranoid: it can confuse dust raised by tractors, insects that pass furtively in front of the camera or simple fog with the start of a fire.
“When a cloud passes, (…) it can cast a shadow on the ground and it can sometimes think that it is smoke,” laughs Suzann Leininger, intelligence specialist at Cal Fire.
Thanks to the expert eye of firefighters and their feedback, the AI continually improves to refine its alerts.
A welcome help, far from frightening the firefighters, unlike many professions – Hollywood actors and screenwriters, accountants, cashiers, etc. — who fear being made unemployed by this technology.
“AI is just another tool for us, it will never replace firefighters,” believes Captain Yeary.
International example
“This saves us time to react more quickly,” adds her colleague, Ms. Leininger. “If we have (…) very strong winds, that can really make the difference between a big fire and a small fire.”
The system promises to be valuable, in a State at the forefront of climate change, which is experiencing increasingly voracious megafires.
California has experienced 18 of its 20 largest wildfires in the past 20 years, and warming caused by humanity’s dependence on fossil fuels “is the driving force behind much of this.” , according to a study from UCLA University published in May.
Beyond the “Golden State”, this pioneering program could serve as an inspiration to firefighters around the world, after the hottest summer ever recorded on the globe, marked by devastating fires in Canada, Europe and Hawaii.
“Given the devastation in Greece and Maui, I think systems like this that provide early confirmation are a step in the right direction,” said Driscoll of the University of San Diego.
Faced with the scale of the threat, the scientist chose to grant public access to the data on his platform, so that other companies or academics can work on it.
“We must leverage all our strengths and work together, because climate extremes are beyond us all,” he concludes.