BILLIONS OF TEXT posts, photos, and videos are uploaded to social media every day, a firehose of information that’s impossible for human moderators to sift through comprehensively. And so companies like Facebook and YouTube have long relied on artificial intelligence to help surface things like spam and pornography.
Something like a white supremacist meme, though, can be more challenging for machines to flag, since the task requires processing several different visual elements at once. Automated systems need to detect and “read” the words that are overlaid on top of the photo, as well as analyze the image itself. Memes are also complicated cultural artifacts, which can be difficult to understand out of context. Despite the challenges they bring, some social platforms are already using AI to analyze memes, including Facebook, which this week shared details about how it uses a tool called Rosetta to analyze photos and videos that contain text.
Facebook says it already uses Rosetta to help automatically detect content that violates things like its hate speech policy. With help from the tool, Facebook also announced this week that it’s expanding its third-party fact checking effort to include photos and videos, not just text-based articles. Rosetta will aid in the process by automatically checking whether images and videos that contain text were previously flagged as false."
So we need to work on how to drive fakebooks AI nuts and cause it to have a mental breakdown!
view the rest of the comments →
CelticCoyote ago
"FACEBOOK'S AI CAN ANALYZE MEMES, BUT CAN IT UNDERSTAND THEM? https://www.wired.com/story/facebook-rosetta-ai-memes/
BILLIONS OF TEXT posts, photos, and videos are uploaded to social media every day, a firehose of information that’s impossible for human moderators to sift through comprehensively. And so companies like Facebook and YouTube have long relied on artificial intelligence to help surface things like spam and pornography.
Something like a white supremacist meme, though, can be more challenging for machines to flag, since the task requires processing several different visual elements at once. Automated systems need to detect and “read” the words that are overlaid on top of the photo, as well as analyze the image itself. Memes are also complicated cultural artifacts, which can be difficult to understand out of context. Despite the challenges they bring, some social platforms are already using AI to analyze memes, including Facebook, which this week shared details about how it uses a tool called Rosetta to analyze photos and videos that contain text.
Facebook says it already uses Rosetta to help automatically detect content that violates things like its hate speech policy. With help from the tool, Facebook also announced this week that it’s expanding its third-party fact checking effort to include photos and videos, not just text-based articles. Rosetta will aid in the process by automatically checking whether images and videos that contain text were previously flagged as false."
So we need to work on how to drive fakebooks AI nuts and cause it to have a mental breakdown!
Triceratography ago
Sarcasm is likely not detectable.