Meet the team using machine learning to help save the world’s bees

Bees have been present in Katharina Schmidt’s family for as long as she can remember – and then some. 'I asked my grandpa once, "How long have we had bees in the family?" He said, "At least since 1894." That was the year his dad was born.' She can still vividly recall the bee hut next to her grandpa’s house – until one year, in the mid 1990s, when it was gone. All the bees had died.

Katharina’s interest in bees reignited three years ago, after she read about global colony collapse. She started her own hive and began helping other beekeepers market their honey – but quickly realised the true scale of the problem. 'The problem isn’t actually that there aren’t enough honey bees. The problem is that bees and other insects are dying around the world in large numbers and we don’t know why. And that was really unsatisfying to me.'

A year and a half ago, Katharina sat down with her roommates to try and figure out a way to help save the bees. Soon after, her company, apic.ai, was born. To collect data, the team constructed a hive monitor using TensorFlow (Google’s open-source machine learning framework). Equipped with a camera that records footage 24/7, it tracks things like how many bees return to the hive every day, how they’re moving and if they’re carrying pollen. The data is then shared with experts so that they can make more informed decisions on things like where to plant trees and flowers.

In Katharina’s eyes, access to information leads to more informed choices. 'Our hope is to provide the grounds for better decisions for the people who are in power. If you show them the data, and you show them the consequences of their actions, you are able to create a changed mindset that will alter the fate of the world, basically.

At least, I am hoping that is what is going to happen.'