Australia’s famed Gold Coast, a 43-mile surfing mecca along the country’s eastern shores, is looking to use AI-powered drones to warn people of sharks, lest they become chum.

The humid, subtropical climate has seen fourteen shark attacks in the last two years that have resulted in two deaths, but a new shark-spotting initiative will officially debut in September after a year of R&D. It involves quad-copters that will fly above the greenish-blue water, relaying video to image-recognition technology that will determine if the footage is of mere dolphins, or something more deadly.

And if a Jaws-like creature is confirmed, the drone sounds an alarm and can drop a four-person life raft and communication device that could enable swimmers to call for help.

Australia’s the Ripper Group is supplying the actual drones, while a team from the University of Technology Sydney developed the AI. The project is being funded by the Australia-based Westpac banking group.

In the near-term, the drones will wirelessly transmit video to a computer on the ground, which will make the decision as to whether there’s actually a shark in the water. In the future, the plan is for drones have to have on-board computing processors that can determine from the air if a shark is in the water.

Little Ripper Life Saver
The "Little Ripper Life Saver" drone will start flying along Australia's Gold Coast next month.

Blumenstein and his fellow researchers essentially devised a program that used publicly available aerial videos, and their own footage, of sharks in the water, to train an algorithm to distinguish between a shark and a surfboard and a swimmer.

It’s just one additional way to keep swimmers safe in the water, Paul Scully-Power, a co-founder of the Little Ripper Group, told Reuters:

“I guess the world has learned many years ago — defense in depth is the way to go. So this is one of the layers of the depth,” said Scully-Power.

When it comes to shark-spotting, humans have only about 18 percent accuracy from a helicopter and just 12 percent accuracy from an airplane. The drones, meanwhile, are accurate 90 percent of the time, reported Digital Trends earlier this year. The project began in 2016 and saw researchers from the University of Technology Sydney capture more than 8,000 images from coastal waters to inform the algorithm.

“It’s not about replacing human beings all together, it’s about assisting human beings to get the work done in a better way with more accuracy. That’s what the application is meant for,” Dr. Nabin Sharma of the University of Technology Sydney told Reuters.

While fear of AI and creeping automation of the human workforce is certainly justified, this is one case where a little extra help from the machines is most welcome.