AI technology is still learning about humanity, but it’s on a similar page as many Americans when it comes to Donald Trump.
A video from a CNN highlights reel of the Republican frontrunner’s presidential campaign announcement shows Trump turning into a literal monster with every flash and instance of raising his voice. It starts off soft, but it soon turns into something much more frightening once the video is run through Google’s Deep Dream generator, which teaches computers how to see and understand the world.
The nightmare-fueled video is courtesy of photographer Eric Cheng, who used audio volume to determine how deep the dream ran throughout Trump’s speech. The louder he was, the deeper you went.
“I ran Deep Dream A.I. against a Trump highlights video (video from CNN, last year),” Cheng wrote on Facebook. “I learned a lot in the process of getting the thing to run fast enough to process video frames without waiting days for results. It’s hard to iterate on this stuff because it is so compute intensive.”
For fun, he added, he used Cthulhu as the guide image for the sequence.
It certainly does the trick.
H/T Gizmodo | Screengrab via Eric Cheng/Vimeo