Why deep fakes are here to stay
Many of the photos and videos out there are already deep fakes.
Slate has a fascinating story about the current state of cellphone camera technologies. Recently, people have been sharing beautiful images of the Northern Lights.
Even people who didn't chase the meterological phenomenon have been found passing other people's pictures as their own.
The author of Slate's article is a meteorologist who has "attempted to see the northern lights dozens, if not hundreds, of times" from Minnesota, "unequivocally one of the best places to see the northern lights without having to travel to Alaska or Iceland or freaking Svalbard."
But he said "almost every photo or video" is deeply misleading. In other words, they are lies. In real life, what one sees is a "faint gray smudge barely visible on the horizon". The colorful photos being passed around are a product of AI technologies, that pick up wavelengths of light that are invisible to the human eye, especially at night time, accentuate them, and depending on what filter is being applied, beautify them.
To say it politely, those are deep fake images.
***
Most of us have expressed outrage at examples of deep fakes, i.e. using deep-learning AI technology to simulate gestures in images, actions in videos, voice in recordings, or words in text, none of which expressed by the persons implicated in those objects. We don't like lawyers who submit fake arguments, politicians who put up videos of opponents saying things they didn't actually say, or scammers who imitate your friend's voice to trick you into sending them money.
All of those examples, and many more, are readily created with today's AI technologies.
There is a reason why deep fakes will not go away. It's because we are generally not too concerned about their harms unless we happen to be the victims. Most of the time, we are not the victims.
Besides, we like some of the deep fakes. Most people I know aren't offended by the fake northern lights photos. They might even argue that the fake ones are more entertaining than the real ones. Similarly, many people use deep fake images on their resumes or applications to present "enhanced" images of themselves to others. They don't see anything wrong with this practice.
The tech industry likes to think that technologies are neutral, while there are good and bad users. This line of thinking has always led to the spread of technologies that are seized by bad actors to do bad things. And so this appears to be the path we're heading down with deep fake AI technology.
