The Future Is Faster Than You Think: How Converging Technologies Are Transforming Business, Industries, and Our Lives (Exponential Technology Series)

by Peter H. Diamandis

In 2018, a YouTube video of former President Barack Obama made its way around the internet. Over 6 million people tuned in to see him seated beside an American flag, speaking earnestly to the camera. “President Trump,” he explained, “is a total and complete dipshit. Now, I would never say these things. At least not in a public address. But, someone else would, someone like Jordan Peele.” Then the video shifts to a split screen. On the left, Obama continues talking. On the right, we see actor-director-comedian Jordan Peele actually speaking the words being put into the former President’s mouth. The video is a deepfake, an AI-driven, human image synthesis technique that takes existing images and videos—say, Obama speaking—and maps them onto source images and video—such as Jordan Peele imitating President Obama insulting President Trump. Peele created the video to illustrate the dangers of deepfakes. He felt the need to make it because it’s really just one of thousands. Political chicanery, revenge porn, celebrity revenge porn—they’ve all been tried and tried again. And while these early versions are pretty close to the genuine article, their fakery is detectable. Researchers at Carnegie Mellon University have recently developed new algorithms capable of far greater realism. Their AI not only transfers head position, facial expression, and eye gaze from one video to another, it transmutes subtle details—eye blink rate, slight eyebrow flicks, the micro-wiggle of a shoulder. An...