Wednesday, April 18, 2018

A Marriage Made In Hell

Now this is really disturbing.

The Verge reports that Jordan Peele and Buzzfeed combined forces to make a fake Public Service Ad:
Using some of the latest AI techniques, Peele ventriloquizes Barack Obama, having him voice his opinion on Black Panther (“Killmonger was right”) and call President Donald Trump “a total and complete dipshit.”

The video was made by Peele’s production company using a combination of old and new technology: Adobe After Effects and the AI face-swapping tool FakeApp. The latter is the most prominent example of how AI can facilitate the creation of photorealistic fake videos.
Researchers have developed tools that let you perform face swaps like the one above in real time; Adobe is creating a “Photoshop for audio” that lets you edit dialogue as easily as a photo; and a Canadian startup named Lyrebird offers a service that lets you fake someone else’s voice with just a few minutes of audio. Technologist Aviv Ovadya summed up the fears created by this tech, asking BuzzFeed News, “What happens when anyone can make it appear as if anything has happened, regardless of whether or not it did?”
The implications of this technology are frightening. Consider, for example, that propagandists will now have a powerful new tool with which to virally undermine their targets with embarrassing or compromising 'videos'; moreover, those who are caught in all manner of malfeasance will, as the current president of the U.S. regularly does, be able to claim it is all "fake news."
Scientists are currently creating tools that can spot AI fakes, but at the moment, the best shield against this sort of misinformation is instilling everyone with a little more media savvy. If you see a provocative video, you should ask yourself: where does this come from? Have other outlets corroborated it? Does it even look real? In the case of AI-generated videos, you can usually see that they’re fake by telltale signs of distortion and blurring.
As always, critical thinking will be paramount. However, how many, no matter how fair and balanced they consider themselves to be, will be able to resist the natural urge to believe the worst about those whose views and practices are so diametrically opposed to their own?

Artificial Intelligence meets fake news: surely a marriage made in Hell.


  1. This is a powerful demonstration of a technology that is already beyond any hope of containing it. It leads to an abyss where minds are manipulated freely by the highest bidder. As we learned from Cambridge Analytica the technology now exists to scrutinize and analyze us and then target us individually with messaging that can both incite and deter us in making choices. Every opposition supporter you can confuse and befuddle is as good as a vote stolen.

    The Obama clip isn't perfect but we're told that advances underway in AI will soon be able to overcome these few rough spots. We've been warned. Stephen Hawking warned us this was coming. So too, I believe, did Elon Musk.

    1. I am currently reading a book entitled The Death of Expertise, Mound. The part I just read about confirmation bias made me think that it, coupled with this powerful new technology, means we have much to worry about.

      Here in Ontario, even the basic campaign of vagueness and Trumpian rhetoric being practised by Doug Ford is apparently yielding him massive support. The future looks rim indeed.

  2. Lying has become so easy. And it's even easier to be anonymous.

    1. Lying has become the default position of far too many, Owen.