Before you interject, I’m NOT going to launch into some tirade about fake news. While we do have to be careful about what news we get on the Internet, it can be just as important to recognize dubious images we see on Facebook or Instagram. That’s where a pair of professors at the University of Washington come in, with the help of some Skynet-level tech to help you weed out fake photos you see online!
Jevin West, who is a professor at UW’s Paul G. Allen School of Computer Science & Engineering, and Carl Bergstrom, who is a biology professor, developed the website whichfaceisreal.com as a tool to test if you can tell the difference between a REAL photograph of an actual human being or a picture generated by a software called StyleGAN.
The program can create an image of a human face within seconds, using “learning algorithms” (whatever that means!) Here’s an example, taken directly from the website where you just click on the image you think is the real one:
If you can’t tell the difference, West & Bergstrom have a tip for you; look for small anomalies that look like “water splotches”. Since the computer isn’t perfect at creating realistic-looking images yet, it leaves these splotches (often near the hairline) as well as failing at creating symmetrical contours in the face, or eyes that seem too creepily identical.
The technology was created in 2017 by workers at NVIDIA, the company behind the fantastic video graphics cards for PCs. The UW professors hope to educate as many people as possible to the fact that just because a profile or an email has a photo, it doesn’t mean it’s real. There’s 21st-Century trust issues that we should all have online when interacting with dating, job postings, emails and the like.
Feel free to play the game yourself! With a little practice you could be better than me, and I have the IQ of a smallmouth bass!
Do you have the 98.9 BULL app yet? You can use it to score free tickets to awesome PNW concerts, and we guarantee no fake stuff. Get it here or text APP to 96459
~Ben On The Bull