Facebook Is Mining Your Instagram Photos to Train Artificial Intelligence

Thursday, 03 May 2018 - 9:38AM
Technology
Science News
Dystopias
No
Thursday, 03 May 2018 - 9:38AM
Facebook Is Mining Your Instagram Photos to Train Artificial Intelligence
< >
Pixabay
Your expertly filtered and hashtagged photos of Fit Tea and hot yoga poses keep your parents updated on your life (even though you never call) – but, as it turns out, that isn't the only service you provide as an Instagram user.

Facebook (which owns Instagram, for those living under a rock) secretly used billions of publicly available photos on the social media platform to teach artificial intelligence software how to better identify objects.

Speaking at the annual F8 developer's conference, chief technology officer Mike Schroepfer revealed the AI project and spoke about how Facebook is using your images – and their hashtags – to make computers smarter. "We rely almost entirely on hand-curated, human-labeled data sets," Schroepfer said, referencing some 3.5 billion pictures used to train the software. They were banking on the fact that people typically use proper descriptive language in captions and tags. Otherwise, the information used to train the AI would be ineffective. "If a person hasn't spent the time to label something specific in an image, even the most advanced computer vision systems won't be able to identify it."

According to Schroepfer, the training has been a success. Facebook hasn't just created a system to rival Iron Man's Jarvis – they've raised the bar to an entirely new level. "We've produced state-of-the-art results that are 1 to 2 percent better than any other system on the ImageNet benchmark," he said.

As Nick Statt of The Verge points out, the project comes at an awkward time in the conversation about privacy. It raises a complicated question of ethics, and whether users know exactly how Facebook uses the data they have uploaded to the various platforms the company owns.

Beyond being able to identify mundane objects automatically, Facebook is developing AI systems to eliminate threats and material that it considers inappropriate. "Until very recently we often had to rely on reactive reports. We had to wait for something bad to be spotted by someone and do something about it," said Schroepfer. "This is why we are so focused on core AI research. We require new breakthroughs, and we require new technologies to solve problems all of us want to solve."

Science
Artificial Intelligence
Technology
Science News
Dystopias
No
No