Alphabet X’s Bellwether harnesses AI to help predict natural disasters



The world is on fire. Quite literally, much of the time. Predicting such disasters before they get out of hand — or better yet, before they happen — will be key to maintaining a reasonable quality of life for the coming century. It’s a big, global issue. It’s also one Alphabet believes it can help tackle.

The Google parent’s moonshot factory X this week officially unveiled Project Bellwether, its latest bid to apply technology to some of our biggest problems. Here that means using AI tools to identify natural disasters like wildfire and flooding as quickly as possible. If implemented correctly, it could be a gamechanger for first responders.

“Bellwether is X’s moonshot to understand and anticipate changes across the planet, so that any organization, community, or business can ask smarter and more timely questions about the natural and built environment,” project head Sarah Russell says in a social media post. “Until now, it’s been epically hard and expensive to apply AI to geospatial questions, but our team has harnessed some of the most recent advances in machine learning (plus some straight-up solid engineering) to re-think the whole endeavor.”

Project Bellwether’s coming out party coincides with news that the United States National Guard’s Defense Innovation Unit (DIU) will be utilizing the organization’s “prediction engine.” According to the teams, current technology has the potential to delay response times by hours or days, causing untold damage to human life and property.

“Right now, our analysts have to spend time sorting through images to find the ones that cover the areas most affected by natural disasters,” the Guard’s Col. Brian McGarry notes. “They then have to correlate those images to surrounding infrastructure, label all the relevant features, and only then can highlight the significant damage and send it forward to first responder teams.”

The Bellwether team has thus far produced two tools. The first is designed to forecast the risk of wildfire “up to five years into the future.” The second is a response tool that helps first responders “identify critical infrastructure” in the wake of natural disaster or extreme weather.

Google has been exploring the use of machine learning models and AI to predict natural disasters for some time now. Project Bellwether’s partnership with the National Guard could well prove an important validation of that work.




Source