Friday, July 28, 2017


I think it’s safe to assume at this point that the least gullible of the American public (which I sincerely hope includes you, Dear Reader) understands that the danger of “fake news” is not that it convinces people of its own veracity (although it can be extraordinarily effective in that regard), but that it degrades truth as a guiding principle in forming your worldview.

That being said, truth is about to start getting degraded a whole helluva lot faster over the next few years, due to a couple of emerging technologies that are seriously goddamned freaky. WNYC’s Radiolab has the scoop:

I’m hard-pressed to say which one has more insidious implications, so we’ll start with VoCo first. A collaboration between Princeton University and Adobe Research, VoCo is audio editing technology that allows you to not just edit pre-recorded spoken audio, but literally to create new spoken audio from whole cloth simply by adding or removing words using a text-based interface.

VoCo works by taking your recorded speech, which it has broken down into thousands of samples based on the phonetic patterns of your voice, and remixing it into new words. What’s more, the new audio sounds all but completely indistinguishable from the original audio, because it is the original audio.

So much for, “I never said that,” eh? Now, that’s merely a technicality.

I’ll get to the rest in a second, but I just want to say that, when I heard this in actual use for the first time, my blood ran cold. Fox News and Breitbart may suck up all the oxygen when it comes to being the face of American clown fascism, but right-wing hate radio is the glue that holds the conservative movement together. If there’s anywhere this technology could do some real damage – and sooner rather than later – it’s in the hands of someone like Rush Limbaugh or Alex Jones or Sean Hannity or the like, especially because of the nearly impervious social membrane that separates hate radio from the mainstream press.

Doctored audio would be able to bubble up from the swamp with ease, attended by completely fabricated talking points that will have maximum credibility in conservative echo chambers everywhere, counterfactuals be damned. When you play them the actual video, they’ll just say it’s fake one. That’s how this works. That’s how this always works.

Up next is Face2Face, software designed to capture facial movements in real time and apply them over existing video:

Technology similar to this is currently in use for doing motion capture work in Hollywood, but the amount of detail and fine motor movement on display here is truly unprecedented. Human expression is one of the hardest things to reproduce with animation, and the fact that it can now be done in real time, however roughly, should have the hairs on the back of your neck standing up right…about…now.

Radiolab integrated the two technologies together to create this clip, and while the technology is clearly still in its infancy, there’s little doubt as to its incredibly disturbing potential. When you combine that with the ability of technology to consistently, drastically outpace its ability to be legislated, there’s little doubt in my mind that, once VoCo and Face2Face are ready to go to market (a question of when, not if), “fake news” as we know it is about to get a whoooooooole lot faker.

No comments:

Post a Comment