Scientists — and surfers — looking to AI to spot sharks

While shark attacks are far rarer than most people’s Jaws-inspired fears, they do happen, and humans are increasingly leveling high technology to keep themselves at a safe distance from the prehistoric creatures. 

While beaches prone to visits from sharks have deployed drones for a bird’s-eye view of the water, sharks can be tough to spot while they’re below the surface. In fact, a study by Australia’s University of New South Wales says some 60% of sharks went unnoticed by drone pilots.

Moreover, it’s tough for human eyes to determine whether that dark splotch in the water is something harmless, like a seal instead of a shark.

To that end, its researchers are marrying Artificial Intelligence to drone footage of sharks to teach computers how best to spot them: The university says AI has proven to have a 90% detection rate.

A drone company called Ripper TechLab has partnered with the University of Technology Sydney (UTS) to develop Spotter AI software, which can be used to not only find sharks from the air, but also crocodiles, dangerous rip currents or swimmers stranded in the water.

Similar tech is being used closer to home, with the University of California Santa Barbara’s Benioff Ocean Science Laboratory’s SharkEye initiative, but researchers both here and abroad are grappling with how to make it available to everybody.