Descending Into The Uncanny Valley: How Our Technology Is Getting Creepier
Updated: Apr 19, 2020
Earlier this week I revealed the front cover for my upcoming sci-fi crime thriller, Auxiliary, which imagines what life in London will be like in two decades' time. In this novel, and also in my short horror story collection Jon Richter's Disturbing Works (Volume One), I explore the dark side of what emerging technological innovations might mean for the human race.
I was inspired to write these books because I am fascinated by some of our recent advancements, particularly in robotics and AI. But why do we find these concepts so intriguing? I think there's something compelling about the feeling of discomfort and unease, even outright revulsion, as our ever-more-realistic machines lead us into the uncanny valley.
But what exactly is meant by the term 'uncanny valley'? The phrase refers the hypothesised dip in a graph that plots how much we like something as it becomes more and more human-like. In other words, we like things the more they resemble us (teddy bears are nicer than waste paper bins, for example), but when they NEARLY resemble us (like a dead body), we suddenly like them a lot less. As shown here, the effect becomes even more pronounced when the item is MOVING... <shudders>
A great example of the uncanny valley effect, albeit a mercifully stationary one, is the website www.thispersondoesnotexist.com. Here, one portion of an 'adversarial' AI generates disturbingly lifelike human faces completely from scratch every time you click the refresh button, with another portion using face recognition software to check and fine-tune them.
At first glance, many of the generated images appear completely realistic, and therefore harmless... but when you look a little closer, or when the system throws up a particularly disturbing anomaly, your skin begins to crawl. Teeth are blurred together, ears are weirdly elongated, eyeballs merged with the surrounding skin. The system seems to have immense difficulty with hats - one of the snapshots I generated featured a man whose grey hair had been styled in the shape of a baseball cap (!?) - and some of the backgrounds are eerily surreal, like a stew of blurred images from which the lifeless face has been conjured. Worst of all is the occasional glitch that spawns grotesque additional faces just off to the side of the main image...
We are already surrounded by robots and AIs, with internet bots and algorithms helping us shop online, and driverless cars beginning to take to our roads. After a recent flight to Iceland, one of the first things I saw when I emerged from the plane was an autonomous robot slowly trundling around the airport, cleaning the floor. All seem very basic compared to our favourite science-fiction mechanoids, but are clearly improving at an exponential rate.
Speaking of those sci-fi stories we were raised on, we are all familiar with robots that are indistinguishable from humans, and indeed most of us probably foresee a world populated by these devices to be inevitable. But what about the midpoint between then and now, the sinister nadir of the uncanny valley, when our machines are a much more familiar in everyday life... but still not quite human? Will we really be able to live comfortably alongside these contraptions?
I think the answer will lie in whether we can cope with the glitches. A construction robot that builds pristine homes at a tenth of the speed and cost of a human crew might be appealing, but what about when something goes awry, and it covers an external wall in toilet seats instead of timber cladding? A lifelike humanoid companion that lives in your house and does your chores sounds great, but what if it starts following you from room to room, muttering nonsense through a manic grin?
Until then, we will continue to speculate, and wonder just how deep this uncanny valley really goes...