Deepfakes

This is a short, fictional piece that I wrote for La Revista de la Universidad de Mexico, as part of their “Diario de la pandemia” series. To read the Spanish version please follow this link.

While we were in the midst of the lockdown in London, and I was sheltering in my house for fear of dying alone in an intensive care unit without enough ventilators to go around, a journalist reached out to me via email for an interview. I found this weird, since I am not an expert in epidemics, or viruses, or indeed ventilators. When I told her so, she ignored me and sent me her questions anyway, which were a list of conspiracy theories about Covid-19. Did the Chinese engineer the virus in a secret laboratory? Was there a link between 5G and the pandemic? How come Bill Gates saw this coming years ago? Was he part of a masterplan by the Illuminati to get everyone vaccinated with a nanochip? And what was Soros’ role in all this? I was asked to offer my opinion.

I really did not know what to answer, as I could not possibly see how my opinion mattered, or indeed how an opinion would be at all relevant. I do not generally believe in conspiracy theories, but then again conspiracies have happened, large and small, throughout history, and will certainly continue taking place as long as people exist. The specific theories that circulated during the pandemic seemed to me exceptionally ridiculous, but there were millions of people who believed in them. So, a better question for the journalist to explore ought to have been why that was. But the journalist was not interested to do a heavy piece, she was writing for a fashion website and just wanted a short, funny, highhearted Q&A with a writer to appear next to an advertorial of a sports apparel brand.

By coincidence (like conspiracies, coincidences are also known to happen) I was writing at the time an article for a popular science magazine on a relatively new technology called “deep fakes”. Essentially, you can nowadays use apps on your phone to create short videos of persons that never existed, make celebrities have sex between them, politicians say whatever you want them to say, and generally amuse yourself and your dozen or so followers on social with fun stuff. Or you can trigger World War Three and Armageddon. Just imagine, for example, Trump losing the 2020 election and a deep fake video have him announcing that the election was rigged and that he does not accept the outcome (same scenario could run with Biden). Or a deep fake video of Kim Jong-Un declaring that he has just launched nukes in the direction of Tokyo. But what is particularly interesting about this terrifying technology for nerds like me is how deep fakes are made using an Artificial Intelligence technique called Generative Adversarial Networks, or GANs for short.

GANs are made up of two artificial neural networks, working against each other. One network constantly creates fake images, starting with white noise. For illustration let’s call that network “Donald Trump” (I am not being partisan here, just bear with me). The other network – call it “Liberals” – takes two inputs: the fake image input from Donald Trump, and a real image from the real world of true reality. Liberals compare those two inputs and call bullshit every time they discover that Trump is feeding them fake images (ok, “fake news” if you prefer). But – and here’s the genius of the system – Trump takes the output of the Liberals’ judgement and uses it to improve the next fake image. Run this back-and-forth dialogue a few thousand times and the Trump network ends up creating fake images that Liberals cannot tell the difference from the real ones. Fake and real are now indistinguishable.

GANs are the ultimate content machines. They can create text, images, music, or videos. I wanted to tell that journalist that her days were numbered, and that soon a GAN was going to replace her, but no need to be mean, right? GANs are also one of Jonathan Swift’s techno-prophecies that has come true.  In Book III of Gulliver’s Travels, Gulliver is abandoned by pirates on the continent of Balnibarbi. After a visit to the flying island of Laputa, Gulliver is taken to the Academy of Lagado, where “useless projects” are undertaken. There, he is given a demonstration of a word machine, a giant mechanical computer used for making sentences and books. The wise men of the Academy pride themselves for discovering a machine that renders obsolete any study or expertise; for now, even an absolute idiot can write a masterpiece by virtue of cranking the machine.  Equipped with GANs, twenty first century idiots can earn the Nobel Prize in Literature. Or run the world from their office via their twitter feed. Or make the world disappear. Or make it burn. To paraphrase Andy Warhol, reality is what you can get away with.

As I was thinking all that, the end of civilization, Pandemic Apocalypse, Trump versus Liberals, Kim Jong-Un, and getting increasingly depressed, I started reading reports from scientists who doubted that the virus was as deadly as it was initially made up to be.  Apparently, the initial mathematical models were wrong. Their code was full of bugs. The government experts had overestimated the death toll of the pandemic, thereby spooking the hapless politicians who went hysterical and turned half the planet into a colossal prison camp. Voices were raised against the lockdowns, protesting that the so-called medicine was worse than the disease. Meanwhile, the number of unemployed skyrocketed. Miles-long lines of cars (many of them expensive-looking) started appearing on newsfeeds, their drivers waiting for hours to scrape groceries from foodbanks.  Two opposing realities competing for domination of the public sphere emerged, like a pair of adversarial GANs creating a deep fake by imitating reality. Scientists were not being helpful either. They are used to brainy debates where multiple truths can co-exist until data and experiment prove most, or all, of them false. But the process of scientific falsification is unknown outside the walls of University campuses, those latter-day Academies of Lagado.  Out there in the wild, in the cities and the countryside, to figure out what’s real and what’s false is too much effort, people have other things to do, like lining up for food and toilet paper, and so they’re more willing to go with their political or tribal affiliations. Thus, the pandemic quickly turned political. There were Leavers (get out, save the economy) and Remainers (stay in, save lives).  If you leaned to the Left you probably went with Remain, if you bent to the Right with Leave. Living in the UK through the trauma of Brexit, I had hoped that we were past that.

As the lockdowns eased across Europe and people emerged from their homes, like snails after the rain, a lot of talk has now turned into how the world will change because of the pandemic. We will fly less, work from home more, use bicycles rather than buses, greet each other with a namaste or an elbow touch, wear masks and gloves, have virtual sex, use apps that track us, wash our hands with soap fifteen times a day. It all feels a little unreal to me. So I emailed the journalist who had asked for my opinion on conspiracies and suggested that I should write an article about the epidemic being a simulation – the deepest of deep fakes – and explore the possibility that we are living inside a hyperintelligent computer that is trying to predict how (the real) humanity may react in a real pandemic. I sent her my proposal by email two weeks ago and have left several messages on her WhatsApp. She has yet to come back to me. Which makes me wonder if she was ever real.