Welcome to the unhappiest place on Earth. Westworld, the futuristic theme park staffed by android cowboys you can murder for fun, is closed for renovations. And a revolution. According to HBO’s trailers for series two of the thoughtful sci-fi, the robotic ‘hosts’ like Dolores (Evan Rachel Wood) have developed full consciousness and are now hunting the humans on horseback. Fans are left asking one big question: what about the horses?
Westworld’s horses are also robots, just in a different shape. So will they become conscious and start stamping on the humans? The bulls certainly seem to have had enough, charging down their human masters in the Superbowl trailer of Westworld season 2, which will be available on Sky Atlantic and Now TV. Plus, what would be the difference between a 'conscious' robot horse and a robot merely simulating a horse?
5,000 years of carefully bred docility have made us forget that horses are dangerous animals. They’re 1,000lbs of muscle and we give them metal shoes. A recent survey found that Australian horses are more deadly than all of the country’s venomous animals combined. Plus, horses remain better able to navigate the land than we are – they could easily run down a fleeing cowboy, robot or not.
Dr Sangbae Kim at MIT is a pioneer in ‘biomimetic’ robotics, which takes inspiration from the natural world. “We figured out how to fly in the sky, move on water, even under the water,” he explains, “but we’ve had to modify our land to use our technology: we had to put down roads or rails. How do we learn from land animals?”
We’ve been replicating horses mechanically since William Brunton’s ‘Steam Horse’ in 1813 – a train pushed forward by two equine legs. In an omen of a potential robot-horse-uprising, its boiler exploded, killing thirteen people in one of history’s first recorded rail disasters. Today, Boston Dynamics regularly thrill and terrify the internet with their four-legged robots designed for the battlefield – they can even open doors now.
MIT’s Biomimetic Robotics Lab has developed a robot ‘Cheetah’ that can run at over 14 mph and jump half a metre in the air. Yet the legs of Kim’s cheetah are nowhere close to the real thing. “You’re comparing about 600 muscles to 12, and when it comes to brain intelligence the ratio is much worse.”
Dr Robert Howe, Professor of Engineering at Harvard, sees a limit to simply copying nature. He points to past failed attempts to build human-like hands, and the later success of ‘underactuated’ hands, with fewer motors and fingers. They are inspired by nature, but a more streamlined solution – horses for courses.
“The human hand has something like 15000 specialized nerve endings,” he explains, “It was like we were trying to build an F1 car before the Model T.”
The point is that under the familiar skin, our robotic horse will likely have the power of a dangerous animal but be fundamentally unlike them: a literal Trojan horse. We might know how regular horses will react, but this is something new and alien, a horse of a different colour altogether. A Boston Dynamics bot won’t eat sugar lumps or frolic in the sunny meadow of your dreams, and it won’t be a product of thousands of years of breeding and training. We might someday be able to build a convincing robot horse, but how do we make it think like one?
In the show, Westworld’s founder Robert Ford (Anthony Hopkins) built the hosts according to the principle of the ‘Bicameral Mind’. Developed in the 1970s by Julian Jaynes, it claims that early humans experienced our own thoughts as the external voice of gods. We were “noble automatons,” not truly conscious until we started telling stories like Homer’s Iliad. For Jaynes consciousness is a matter of ‘narratization’: a story we tell about ourselves, to ourselves, to explain our actions.
It’s a trippy idea and perfect fodder for a show that is symbolically about our own violent culture getting revenge. But unlike humans and hosts, horses don’t speak and aren’t motivated by tragedy, unless Dolores is going to start riding Bojack Horseman. Without language, horses are inscrutable. We don’t know what they are thinking, or how they think, or even if they think. Looking into those watery eyes is to glimpse a brain unlike ours. And this difference has justified horrors before.
Descartes, the father of modern philosophy, declared that the only certainty is “I think therefore I am”. He was less sure about horses. For Descartes consciousness was fundamentally a matter of thinking, a position known as conceptualism. Animals can’t talk; therefore they can’t think; therefore they’re not conscious.
Instead he viewed animals as a kind of biological robot, mindlessly reacting by instinct. “Men can make various automata which move without thought,” he wrote, so it seemed reasonable that “nature should produce its own automata much more splendid than the artificial ones. These natural automata are the animals.” In Descartes’ view, we already have robot horses: they’re standing in fields right now.
As the hosts will tell you, humans don’t treat robots well, and that goes for ‘natural automata’ too. Descartes took part in horrific vivisections and, centuries before Pavlov, mused that whipping a dog to the sound of the violin would make it “run away as soon as it heard that music again.” Yet even if we find such cruelty abhorrent, can we explain the difference between a human, a horse, a robot and a robot imitating a horse? And what if, like the hosts, they remember the pain they’ve endured at our hands?
Read more: We need to talk about everything that's wrong with Ready Player One
There is no one test for determining consciousness, animal consciousness or true horse-robot-consciousness. For instance the Turing test for A.I. relies on language – robots must pass themselves off as human in conversation to clear it – but we’re getting nothing straight from a robot horse’s mouth. Then there’s the ‘mirror test’ designed by (honestly) Gordon Gallup Jr, that paints dots on animals to see if they recognise their own reflection. It’s a test of self-awareness, yet a pilot study with horses was inconclusive. (It’s worth noting that young humans fail.) In fact, although the ‘higher order consciousness’ of humans is ostensibly a more complex mental state, we seem to find it easier to describe, and easier to verify, than the wordless experience of animals. Our robot horse bolts through the gaps in our understanding.
“It seems easier to answer how do I give a robot higher order functioning” says Dr Michael Tye, professor of philosophy and author of Tense Bees and Shell-Shocked Crabs: Are Animals Conscious, “than it is to answer how do I give a robot basic subjective experiences like pain?”
In fact Tye claims that the only real evidence we have for whether animals are conscious is their behaviour. An animal that acts similarly to us when we are hurt is probably feeling pain, according to Newton’s rule that similar effects should be credited to similar causes.
“It’s the simplest explanation for what’s going on, the best way to proceed with respect to other animals or robots or whatever,” he argues. This means that giving our robot horse ‘consciousness’ would just be a matter of complexity. “If a robot horse’s general range of responses are indistinguishable from real horses, then we have reason to suppose the same experiences and feelings are present.”
Westworld’s robot horses would ‘achieve’ consciousness once it convincingly walked like a horse and neighed like a horse, even if they were built differently inside. Tye emphasises this isn’t ‘proof’ as “I don’t think there’s any chance of getting a proof. But behaviour is the best evidence we’ve got that creatures are subject to conscious states.”
So what would be the difference between a horse and a robot horse standing in a field?
“Only the hardware.”
This suggests some fun philosophical horseplay, and implies some safety tips. For one, it radically expands our definition of horsedom. Imagine our robot horse wasn’t run by a neat computer in its head, but by thousands of tiny Greek soldiers working together, each acting as a ‘neuron’ to imitate a brain. (This is a version of a problem known as the China Body.) Would the sum total of this Trojan-consciousness be a horse too?
It also suggests that Westworld might one day come true – at least in terms of horses. Disneyland has been at the forefront of lifelike robotics since the singing birds of the Enchanted Tiki Room in 1963. Disney Research has labs across the world and, in February, launched the Vyloo - autonomous robots in the forms of aliens. Their robots aren’t just a solution to locomotion like MIT’s Cheetah, but aim to “endow robotic characters with life-like features,” according to their website.
“The engineering that goes into those animatronics is astounding,” says Howe, “designed to trigger all the connections you get when interacting with a real animal or person. You can imagine they'd like to deploy robot Mickeys in the Magic Kingdom.”
Or, say, robot horses in Frontierland. Among Disney Research’s papers are studies into making four legged robots walk.
“I would not be surprised, they would be the ones to make it work,” says Howe. But what about the safety concerns of unleashing them in a theme park? “Like with self-driving cars, with any kind of a complex system it's probably impossible to guarantee complete safety.”
Read more: 20 of the best TV box sets on Sky and Now TV right now
It would seem that the best course of action, both in creating our robot horses and in keeping them from going feral, is to follow Descartes’ lead and treat robot horses as we would any other animal. A truly ‘conscious’ robot horse is one that acts like a horse, and a horse doesn’t have in-built safeguards. You wouldn’t just unleash an unsupervised pack of horses (robot or otherwise) into the middle of Disney World, to careen through It’s a Small World or stamp the Mad Hatter’s Teacups into shards. Neither would we be surprised if an abused horse tried to kick our head in – Descartes’ dog was as likely to bite him on hearing that violin as it was to whine along to the tune. Treating creatures that seem to feel pain as if they feel pain isn’t only a sensible method of judging consciousness, but a version of Kant’s Categorical Imperative – an ethical outlook that might just save our lives too.
And the wider point is that while fans are champing at the bit to continue the story of Westworld’s hosts, it’s their steeds that are the park’s true stalking horses, an encapsulation of its themes. When it comes to consciousness – artificial or otherwise – we should remember that not everything looks or thinks like us. Both the hosts and the humans should get off our high horses.
This article was originally published by WIRED UK
