‘Fake news’ is a term heard from the White House to our Facebook feeds. A new piece of technology being developed in the Capital Region is trying to determine what's real and what's not.
University at Albany professor Siwei Lyu says distortion to pictures in the media has happening for decades with Photoshop. However, nowadays most people get their news from cell phones. The technology "Deep Learning" has been creating realistic-looking fake videos. According to Lyu, it could cause a problem, especially during election season.
"Social media is a total mistrust of information. Now we have a lot of information but we can't trust them, so it's as bad as we have no information," he said.
Lyu says his team is finding ways to crack the code with a special algorithm that finds faults in fake videos. The trick, Lyu says, is in the blink of an eye.
"If you put those images on the internet and people only put what we call 'good images' on the internet. ‘Good images’ usually correspond with images with eyes open, so the AI algorithm usually rarely sees images with eyes closed or eyes blinking,” said Lyu.
He says the technology will continue to advance to a point where these videos may be undetectable. Until then, he says, researchers will have to stay ahead of the curve, and make sure consumers know to scroll with caution.