Putting Words in Your Mouth, Expressions on Your Face

Putting Words in Your Mouth, Expressions on Your Face

Trust, Technology & Artificial Intelligence

 65% of Americans do not trust their government to do the right thing. A huge reversal from three quarters of a century ago, with most of the erosion coming in the last quarter century. 70% of Republicans still do not trust that their state and federal governments delivered a free and fair election last November, such that Donald Trump was in fact the duly elected President… despite lacking a shred of meaningful evidence to support that claim, way over a year later… with minions pouring over whatever they can find. Dems call that the “Big Lie,” and statistics and legal investigations seem overwhelmingly to support that description. 

When right-wing media dramatically accused voting machine manufacturers of hooking their stations, via the Internet, to foreign controlled and pro-Biden sources, their lawyers forced massive retractions when it was proven that these machines are never connected to the Internet. Defamation actions are pending.

More recent moments in the sun for right-wing media? During his term in office, President Trump appointed special counsel, John Dunham (still in the Dept of Justice), to investigate the Russian investigation. Trumpy pundits recently layered in a few embellishments on Dunham’s work. They claimed Dunham had solid proof that Hillary Clinton personally pushed the nation’s intelligence agencies to spy on President Trump. To hear Fox News – particularly stalwart anchors like Sean Hannity and Tucker Carlson who accusedly uttered Hillary’s name literally hundreds of times – and its right-wing mini-networks (OAN and Newsmax), the proof was overwhelming. 

Except on Thursday, February 17th, Dunham distanced himself from the fray, and explained that his investigation had nothing to do with spying on President Trump; he was only investigating what had occurred during the Obama administration. All these news outlets fell embarrassingly silent almost immediately. Hillary mumbled that perhaps all these actions had risen to “actual malice,” the threshold to sustain any defamation litigation against a public news network or publication.

Even though these days the most significant and massive political mendacity seems to come from the Trump-dominated Republican Party, the Dems are hardly above embellishment and fabrication when the mood suits them. And if political parties can lie at this magnitude of malignant fabrication based on innuendo, conspiracy theories and embellishment… without “proof”… what happens when they can manufacture “proof” (or a damned good semblance thereof) that becomes exceptionally difficult to refute. Particularly for those who form their opinions on first impression… when that “impression” jibes with what they want to believe.

Every couple of months, I check into the world of computer-generated audio-video “realistic” simulations. It still takes a bit of expertise and sophisticated software to do it to a level of virtual undetectability, but increasingly off-the-shelf programs, available to anybody, are coming very close to the level of resolution and vocal replication of film and television professionals. And if you want to be afraid, even the title of a report (accepted for publication on December 20th) in the Proceedings of the National Academy of Sciences for the United States should terrify you: AI-synthesized faces are indistinguishable from real faces and more trustworthy by Sophie J. Nightingale and Hany Farid (an academic psychologist and comp sci engineer).

Based on an extensive series of recognition experiments, bottom line, test subjects in this study often preferred the synthetic persona as “real” and found the genuine audio-visual representations as “fake.” The authors warn us: “Artificial intelligence (AI)–synthesized text, audio, image, and video are being weaponized for the purposes of nonconsensual intimate imagery, financial fraud, and disinformation campaigns. Our evaluation of the photorealism of AI-synthesized faces indicates that synthesis engines have passed through the uncanny valley and are capable of creating faces that are indistinguishable—and more trustworthy—than real faces.

“Artificial intelligence (AI)–powered audio, image, and video synthesis—so-called deep fakes—has democratized access to previously exclusive Hollywood-grade, special effects technology. From synthesizing speech in anyone’s voice (1) to synthesizing an image of a fictional person (2) and swapping one person’s identity with another or altering what they are saying in a video (3), AI-synthesized content holds the power to entertain but also deceive.

“Generative adversarial networks (GANs) are popular mechanisms for synthesizing content. A GAN pits two neural networks—a generator and discriminator—against each other. To synthesize an image of a fictional person, the generator starts with a random array of pixels and iteratively learns to synthesize a realistic face. On each iteration, the discriminator learns to distinguish the synthesized face from a corpus of real faces; if the synthesized face is distinguishable from the real faces, then the discriminator penalizes the generator. Over multiple iterations, the generator learns to synthesize increasingly more realistic faces until the discriminator is unable to distinguish it from real faces…

“Perhaps most pernicious is the consequence that, in a digital world in which any image or video can be faked, the authenticity of any inconvenient or unwelcome recording can be called into question… Although progress has been made in developing automatic techniques to detect deep-fake content current techniques are not efficient or accurate enough to contend with the torrent of daily uploads… The average consumer of online content, therefore, must contend with sorting out the real from the fake.” 

In a world of mass political deception, the proclivity of once “trusted” political leaders and major media figures knowingly to lie and accept the falsehoods of untrustworthy sources, imbuing them with their imprimatur of “authenticity,” the big and very unanswered question remains: can a “free speech” democracy survive? Replay the siege on the Michigan State Capitol on April 30th of last year, the Unite the Right Charlottesville, Virginia protest in August of 2017 and last, but hardly least, the January 6, 2021 attack on the US Capitol attempting to reverse election results by violence… and reach your own conclusion. The above sequential photographs are real people reacting to fake news. Imagine how these people could justify their violence if they had “graphic proof” in support of their pernicious views.

I’m Peter Dekom, and every day we learn a little bit more about the fragility of the American experiment with Democracy… which is facing so many existential threats.