American Scientist

The ocean and the amazing creatures that call it home face many threats, and to know if our solutions to these threats are working, we need data. “The biggest challenge that our community faces is monitoring the ocean—critically, we need to not just see things one time, but see them repeatedly and understand how they’re changing,” said Kakani Katija, a bioengineer at the Monterey Bay Aquarium Research Institute (MBARI) in California. “The ocean represents 97 percent of the habitable ecosystems on the planet, and we have long-term observations and monitoring for less than 1 percent of it, especially at a time when we need to monitor these systems as they’re changing rapidly because of climate change.”

Something that may be surprising is that the limiting factor in large-scale biodiversity monitoring isn’t just collecting the data, but analyzing it. There are countless hours of audio and video footage and countless more still images, all collected by scientific instruments throughout the world’s oceans. But only a handful of people are trained in how to analyze them reliably. “The leap from data collection to actionable information is quite a bit,” Katija said. “Only so many people in the world can recognize the call of a particular dolphin species, or look at a particular image of a squid and identify it to the species level. We need to scale our community’s expertise.”

That’s where a new MBARI-designed game called FathomVerse comes into play. FathomVerse trains users to identify deep-sea species from real videos collected by scientists, and then turns them loose to analyze data that no one has ever looked at before. Users simply identify what species are found in their assigned videos, and when several trained users independently agree on an identification, it gets submitted for approval. In doing so, players are helping to train an AI that will help scientists monitor the ocean.

“When you download the game, your goal is to be one of the first people to ever see an image of a species that no one has ever seen before,” Katija said. “But in order to get to see those images, you first have to be trained. There are dozens of training missions in the game, and as you play them, you see real life imagery of animals and learn about them. Once you’ve done a couple of these training missions, you get certifications, and then you can look at raw expedition footage that no other human being in the world has ever seen." Katija thinks the game can compete with other popular apps. As she joked: "These animals are more interesting than Pokémon, fight me!”

So far, players in more than 100 countries have generated more than 3 million annotations for ocean imagery, a wonderful success for this gamified community science initiative. “The game is a lot of fun, but we’re not just collecting data for fun,” Katija said. “We’re collecting the data to help understand the ocean during a time when there’s so much pressure on these threatened species and ecosystems. If we’re going to find solutions that work, we need to understand changes to those communities, and we need monitoring to do that. Playing this game is something that you can do to help.”

To play FathomVerse, visit their website and download the app: https://www.fathomverse.game/