Tech vs. Nature: The Complex Role of AI in Wildfire Control

The recent wildfires in Los Angeles County have highlighted the complex role of artificial intelligence in both fighting fires and potentially contributing to the conditions that fuel them. AI has emerged as a promising tool for wildfire prediction and resource allocation, but its substantial energy and water needs raise valid concerns in a state already facing water scarcity.

This is not the first time the state has suffered wildfires, it has a history of both natural ignitions and human-caused fires. However, over the past few decades, a concerning trend has emerged where wildfires are becoming larger and more destructive.

The Palisades fires are considered the most destructive in Los Angeles County history. According to analysts, the projected damages are reaching tens of billions of dollars, The reports have confirmed that over 20,000 acres have been burned with more than 9,000 structures damaged or destroyed. In addition, over 150,000 people were evacuated in the Los Angeles area.

In response to the 2022 wildfires, Governor Gavin Newsom announced new initiatives to combat future blazes. “We’re enlisting cutting-edge technology in our efforts to fight wildfires, exploring how innovations like artificial intelligence can help us identify threats quicker and deploy resources smarter,” Newsom explained.

One of the initiatives was to train an AI model to analyze video feeds for early detection of fires. If the model identified a threat, it was programmed to alert a human team, who could take action and extinguish the blaze before it became a widespread disaster.

Using this AI model, CAL FIRE and the University of California San Diego’s (UCSD) ALERTCalifornia launched an implementation trial deploying a 24-hour surveillance network of 1,032 pan-tilt-zoom HD cameras for efficient monitoring of active wildfires and other disasters. The cameras were installed in Los Angeles, Santa Barbara, Madera, and other counties.

Were the AI-powered cameras successful in containing the Palisades fires? Well, it’s challenging to assess the impact because it's difficult to determine how many potential ignitions were prevented or how much spread was mitigated by early intervention.

However, if conditions are against you, as they were with the Palisades fires, AI is no match for Mother Nature. The fires spread so rapidly that the AI-powered cameras were not as useful as you would imagine.

“All fires start out as small fires, but when it’s pushed by 60-100 mile per hour winds, and the fuel switches from grass and brush to houses, filled with petroleum products, that’s just untenable.” shared David Acuña, spokesperson, Cal Fire.

California’s AI wildfire detection system has had some success. In December, the Orange County Fire Authority (OCFA) used it to quickly detect and contain a fire in Black Star Canyon. This early detection limited the fire to under a quarter acre.

AI has been used by several other projects to combat California wildfires. Last year, researchers from the University of Southern California (USC) developed a new AI model that uses high-resolution satellite imagery to accurately predict wildfire spread.

The model combines satellite imagery with a sophisticated GenAI algorithm (cWGAN) to forecast a fire's likely path, intensity, and growth rate. Trained on historical wildfire data, the model performed well in tests using real California wildfire data from 2020-2022.

Factors such as topography and weather also influence fire behavior making it a highly complex and nonlinear process. This makes the performance of wildfire AI models impressive. However, these systems are limited to early detection and warning. After all, they can’t change the Santa Ana winds or prevent the dry conditions that fuel these devastating blazes.

Beyond these limitations, the massive data centers powering AI raises further environmental concerns, particularly regarding their substantial water and energy demands. Big tech companies are spending billions to build AI data centers to train and serve large models.

Southern California, which is often the location of wildfires, is also a hub for the AI boom. The region has seen a surge in AI data center energy consumption, resulting in immense strain on the state’s resources. The International Energy Agency (IEA) reported that data centers, globally, used 2% of all electricity in 2022 and the IEA predicted that could more than double by 2026.

The AI data center requires millions of gallons of water for cooling. A single large data center can consume as much water as a town of 50,000 people. This staggering water usage directly impacts the availability of water for firefighting. This is particularly concerning during prolonged droughts and wildfires.

So why does AI need water? They need water to prevent the computers from overheating and malfunctioning. These systems often involve cooling towers that evaporate water to dissipate heat, much like how sweating cools the human body.

Unless we come up with a more energy-efficient way to manage AI data centers, we will continue to consume precious resources. It’s quite ironic that the very technology that we look upon to help us prevent and contain wildfires could be contributing to making it worse.

It’s encouraging to see some companies take action on this. Microsoft is set to begin piloting zero-water technology at new data center sites in Arizona and Wisconsin by 2026. As the trend toward sustainable practices continues, we can hope to save finite resources for where they are needed the most.