Deepfakes and Train Myths

We naturally believe in videos, as the saying “a picture is worth more than a thousand words” is rapidly fading away. Yet it seems that sooner rather than later, we won’t be able to trust videos anymore. Everyone is going nuts about deepfakes on social platforms. But is it the first time we have a hard time differentiating reality? The recently released movie Mountainhead, about which I have my reservations, highlights a tale many film fans take as true. In 1896, a film called Arrival of a Train at La Ciotat premiered, staging the arrival of a train from the perspective of somebody waiting on the platform, standing close to the tracks. The simplicity yet vivid experience allegedly made viewers run out of the cinema, believing a train was going to hit them. With AI, we are starting to see the same thing on social platforms, where some people are not recognizing what’s true or not.

So where do we draw the line? It unavoidably seems that in the near future, without regulation, deepfakes will get so good the naked eye will be unable to distinguish them from a real video. In the book AI 2041 by Kai-Fu Lee, there is a chapter called “Deepfake,” where a programmer in Nigeria is forced to make a deepfake to take down a politician. The story hits particularly hard because it’s not just sci-fi—it’s where we’re headed.  By 2041, like AI 2041 warns, we might not know what’s real unless we’ve got some crazy tech to check every clip. Or my naive option of some sort of watermark to identify them.