Black Mirror tech: how real is the AI in the show?
Credit: Netflix
How far away are we from uploading ourselves to the cloud? We asked AI expert Ian Rowan to break down five uses of AI in Netflix’s Black Mirror.
No television series in recent years has shone a light on artificial intelligence like Netflix’s Black Mirror. The chilling anthology series takes a look at the darker side of tech and how it reflects back on us – hence the show’s title, which is inspired by how a phone screen looks like a black mirror when it’s not unlocked – with much of the technology itself being based on AI algorithms of the real world.
To the casual viewer, writer and showrunner Charlie Brooker’s uncanny ability to predict the future only adds to the ghastly drama: see MIT’s Fifteen Million Merits-style wearable electrode for generating personal power, or Nosedive’s face recognition on phones, later copied by Apple. If it’s prophesied in a dystopian semi-sci-fi show and hits shelves six months later, there’s fair cause for concern about some of Brookers’ more sinister undertones.
With the rate that artificial intelligence is progressing, however, perhaps the show’s mirroring of reality, technology-wise, shouldn’t come as too much of a twist. Life imitating art in this way is perfectly logical with the rate that technology is unfolding. So long as autonomous vehicles are developing, for example, it seems sensible that delivery firms should trial this first. Yet when Pizza Hut announced a delivery vehicle staggeringly similar to the one seen in Season 4 episode Crocodile, it seemed like a big coincidence or some vision of the future.
Black Mirror’s sensible and considered approach to foreshadowing technological advances certainly begs a bigger question. What about the more outlandish technological advances?
Even the idea that you could unlock a mobile phone with nothing more than your face would have seemed absurd two decades ago. At the speed that AI is hurtling in, is it beyond the realms of possibility that we will be able to upload human consciousness to an 80s-tinted utopia after the body of that person has died? With the possibilities of gaming, how real could we make VR? What about the advancements being made in chips that are inserted directly into the brain?
We spoke to machine learning expert and the inventor of Nurio, Ian Rowan about how soon a Black Mirror future really is.
Hang the DJ: Dating app simulations
Featured in the fourth season of the show, Hang the DJ is an episode featuring a couple using a dating app. The twist at the episode’s finale reveals that both parties have had their consciousnesses copied as an AI. The app has been running a thousand simulations of the couple in a challenging world to test how compatible they really are.
“As an AI/machine learning practitioner, I can say that Hang the DJ actually does an incredible job of describing the machine learning process that occurs in many of our apps outside of the deep technicalities,” says Ian. “The actual process we use when training models is very similar to the concept in the speed dating realm of the app in the episode.”
“Many scenarios for any kind of data are iteratively tried and models adjusted in most machine learning apps. Likewise, for each date in Hang the DJ, the outcome is evaluated and the next iteration updated based on these results. Eventually, a training stop condition will be met, in [the case of the story] the characters attempting to run from the overarching force in their simulation.”
Many dating apps from the real world test compatibility. OKCupid, for example, asks questions to users and generates a percentage of how much you have in common with another person based on your answers. Ian suggests that a full brain scan of a dating app user, as implied in Hang the DJ, would be ideal, though perhaps not even necessary given the data that we submit to these companies.
“While a brain scan is likely a dream of companies like Tinder and Match, they are building essentially identical applications as we speak, using more simple data.”
Be Right Back: Chatbots based on deceased relatives
Be Right Back, an episode from the second season of the show, is one centred around grief. After her social media-obsessed boyfriend Ash dies in a car accident, Martha approaches a company to essentially create a chatbot based on the online interactions he made when he was alive.
“This is most likely one of the technologies which is coming quicker to the real world than even the Black Mirror producers likely expected,” says Ian. He himself has trained an AI on his own writing and had fascinating results.
I could likely train a chatbot based on a friend or family [member] any time in a matter of hours!
Ian Rowan, CEO of Nurio
“From my experimentation with this model I was able to train a model to write nearly identical but yet still unique to my style in just a few hours,” he says. “Something like this can easily be adapted to a chatbot to allow loved ones to chat with you whenever they would like. The more data that can be saved from conversations between loved ones, the more possible preserving a chatbot version of them will become.”
So could Ian realistically create a chatbot based on your deceased relative? He claims so. “I could likely train a chatbot based on a friend or family [member] any time in a matter of hours,” he says.
In Be Right Back, the technology doesn’t end with the chatbot. Rather like how internet dating usually progresses from texting to phoning before real-life interactions, Ash’s voice is recreated next for Martha to have phone conversations with before his likeness is reimagined into a life-sized android. This kind of technology is still a long way off, according to Ian.
“Roboticists still have quite a way to go before catching up to the realism displayed by this bot,” he says. “AI research is more or less sitting waiting for this to come with the AI models for the brains becoming more and more ready.”
“Open AI recently released results and models for their GPT-2 language model. This allows a deep neural network architecture to learn to mimic provided writing given any prompt. The overall industry is beginning to trend towards generative chatbots which will allow a near Turing-test passing experience in the very near future.”
Playtest: VR gaming that finds our darkest fears
One of the most horror-inspired stories of the show’s canon, Playtest tells the story of Cooper, who tests a virtual reality gaming headset. The headset plunges the user into a simulated world around them that feels realistic, showing them their darkest fears.
“As far as the technology goes, the overarching theme of this episode is one of Black Mirror’s favourite’s: brain level VR,” says Ian. “We see this similarly in [episodes] USS Callister and Striking Vipers. The concept here is actually something that the neuroscience community is working hard towards: mapping the full human brain.”
This concept is still a long way off. Scientists are getting closer to being able to read signals from the brain, however, our understanding of how it works is still hazy.
“The concept in Black Mirror is a neural understanding that allows these devices to write to the brain, thus creating signal such as optical signals [immersing] the brain in a virtual world,” says Ian. “In theory, if a brain sat standalone from a human body and received identical signals from all of the senses, it would likely still perceive a completely real world.”
“[The Black Mirror Christmas special] White Christmas does a good job of showcasing this theory, as they inject the [AI that they have created from] brain mapping into a home virtual assistant without the consciousness even realising it no longer has a human body.”
Whilst we currently have virtual reality gaming, it still isn’t powerful enough to convince a user that they are in a completely new world. However, one of the aspects that makes stories such as Playtest and USS Callister so horrifying isn’t just how real these games feel, but that the user is being hacked in the outside world. Ian believes that Black Mirror creator Charlie Brooker was particularly influenced by 2010 movie, Inception, when he wrote Playtest.
“As we see in Playtest, Cooper drifts into new games within his games which increasingly slow his perception of time, which is analogous to the dreams within dreams of Inception,” says Ian. “In both of these stories, the dream/VR worlds are completely controlled by a greater party, which is in essence hijacking the brain’s perceptions and senses while the user is more or less asleep.”
San Junipero: A world for consciousnesses to escape death
One of the most celebrated and surprisingly hopeful episodes in Black Mirror’s dark back catalogue, San Junipero is a tale of two people meeting in a simulated world.
In the episode, the technology gives elderly people the chance to have their consciousness extracted from their brain – similarly to storylines in White Christmas and Hang The DJ – and placed into a simulation. The simulation appears to have pain thresholds and feels completely real to those in the world. Subsequent episode Black Museum later describes the technology as “uploading old people to the cloud”.
“Buried deep inside of every brain is a completely unique and incredibly complex pattern of synapse weights and connections which in essence create our consciousness,” says Ian. “As mentioned before, if every piece of the human brain is preserved and the correct signals can be provided, a world like San Junipero could be experienced by any human brain.”
One of the most prominent themes of Black Mirror, San Junipero suggests that human consciousness can be recreated in computer code. This is something that has long been theorised in science and Ian believes that mapping the human brain is the first step. There is still a lot that science does not understand about the inner workings of the brain, after all; should science unlock all of its potential, however, there’s no reason that it couldn’t recreate a person’s consciousness in a virtual world. It would be complex, though.
“In order to construct this world, that level of AI would be necessary to extract the type of interests and fears the person would like to see in their new world,” says Ian.
Crocodile: A screen that replays memories
Crocodile explores memory, touching upon its subjectivity and a person’s ability to recall events that they have witnessed.
In the episode, an insurance company has a device which can watch a user’s memories and display them as videos on a screen. Though this seems like pure fantasy – especially compared to Season 1 episode The Entire History of You, which explores the idea of characters recording their memories via contact lenses to view later – Ian claims that this is another technology that could just be around the corner.
“There are current neuroscience studies which show a user an image and record their optical signals at the same time to create labelled data sets,” says Ian. “Generative Adversarial Networks (GANs) are then used to use the signal to generate the image in the data set. The resulting network can now take in new optical signals and produce its best guess of the image associated. The results have so far been better than expected with new images showing very similar shapes but still not that much detail.”
Memories seem nebulous. The idea that technology could display things that we’ve seen, simply from our recall, may seem a little far-fetched; Ian does, in fact, believe that should the technology come to fruition, it would rely more on our imagination than a strictly factual retelling of past events.
“I think that one issue that Black Mirror ignored in this sense is our ability to force optical thoughts that overcome our memories,” he says. “We can quite easily doctor our memories however we see fit, that’s what we call imagination. Also, the Black Mirror use case would have had to deal with quite a bit of interference with the actual current optical signals from the user’s eyes.”
Like much of the technology within the show though, the machine in Crocodile isn’t a million miles away from the truth. In fact, this is what makes Black Mirror so gripping for the audience. Black Mirror is rarely distantly futuristic: it exists in the technology that we’ve already let into our lives. Many of the concepts are simply expanded versions of things that we already know intimately, such as iPhones, virtual reality and social media.
The best science fiction tells stories that resonate with an audience in some way, reflecting back at them something that they recognise in their contemporary lives. The Daleks in Doctor Who, for example, were intended to mimic the Nazis, a villain that resonated with a 1960s audience who had only experienced World War Two twenty years prior.
Black Mirror is more subtle than to use screaming aliens to recall memories of real-life terrors. Technology is never the evil in the story. This isn’t a TV series set on foreign worlds or spaceships – at least not real ones – but in our phones, in our homes and in our heads. This is perhaps the biggest misconception of Black Mirror: it’s not a series dedicated to showcasing the perils of tech: it’s a show that uses technology to highlight the horrors of humanity.
According to Ian Rowan, it’s not just automated pizza delivery and political commentary from the show that we should be watching out for. Much of the AI used in Black Mirror is real. With the rate that it is evolving too, it might not be too long before the show seems a little outdated compared with the gadgets and services we have in our lives. Is Ian hopeful for a future of technology like that seen in Black Mirror?
“Yes most definitely,” he says. “But it’s a good idea to take note of how bad some of the tech can turn out and avoid those directions.”