Computer brains in person-shaped jars

They serve little purpose other than to freak us out. Why are we so obsessed with creating humanoid robots?

 
 Illustration by  Sarah Nagorcka

Illustration by Sarah Nagorcka

 

This is an editorial for Issue 15 by Lateral editor-in-chief Jack Scanlan, who has never really met a robot he properly clicked with.

Sometime in the coming century, many futurists and prominent science fiction enthusiasts predict, humanity will be eclipsed by non-human minds of our own making. Whether or not this eclipsing is good, bad or neutral depends on who you ask — futurists have never been particularly accurate when it comes to the actual future, but their predictions nevertheless alway spark lively debate and often reflect societal hopes and fears.

When you think of computer-based minds, what do you envision: brains made of software holed up in computers, or walking (and maybe even talking) human-shaped robots? Our society’s near-dependence on the Internet in recent years has made it a lot easier to imagine artificial intelligence (AI) existing and thriving without the need for robotic bodies to inhabit — there’s a whole World Wide Web to explore. But the allure of humanoid robots is hard to ignore, and we seem to crave bringing them into being, both in fiction and reality.

At a basic level, a robot is a computerised machine that can move itself without direct mechanical involvement from a person. Our smartphones aren’t considered robots because they don’t move, and neither are cars because they can’t act by themselves. Self-driving cars fulfil both criteria, but it’s still difficult to consider them robots — most robots are somewhat designed from the ground up, while they seem more like existing technology (the classic car we all know and love) into which a computer mind has been placed.

Of course, many robots aren’t strictly designed from the ground up, and instead mimic living creatures. This has its pros and cons. Some animals can be pretty good at moving about, utilising engineering principles unconsciously discovered by millions of years of evolutionary pressures — many of which human engineers have only just started to understand (or steal for themselves). The self-correcting gait of many animals (notably implemented by Boston Dynamics’ dog-like robot Spot) is a necessity for walking robots, which are vulnerable to the simple act of being pushed over, while bird wings are being mined for their aerodynamic qualities.

But mimicry isn’t always the best move. If you want to build a small, flying robot, should you mimic how an eagle flies, or create something more like a quadcopter? The latter’s rotating wing motion produces lift more efficiently than the up-and-down motion used by birds, and unless your robot is very small, like a hummingbird, hovering in place is next to impossible if you copy nature. Going our own way can save us being trapped by obstacles the blind processes of evolution were never able to surmount. Copying what already exists isn't a guarantee of success. 

So, here’s the big question I want to put forward: why are we obsessed with creating humanoid robots? This applies to the real world as well as our written and televised fantasies. The human body isn’t particularly well-suited for doing most things we might want a robot to do — it's easy to imagine cleverly designed, multipurpose forms with dozens of appendages being a lot more useful. 

The first response might be that creating a convincingly humanoid robot would be a display of society’s robotics prowess. Science fiction often assumes our ability to create more and more lifelike robots scales with our technological advancement, such that if an author jumps into a future where robots indistinguishable from humans wander around like everything’s normal, the audience will implicitly make the leap that this future’s technology far surpasses our own.

In the real world, creating humanoid robots tends to get a lot of coverage in the press. ASIMO became world-famous when it was revealed to the public in the early 2000s, as a benchmark for how far humanity had come shaping the world to its image. We seem to like being reminded that we've come a long way since the steam engine — and even better if the reminder affirms our physical, as well as intellectual, properties.

But it’s one thing for a robot to have a humanoid shape and another for it to look convincingly human. This affects the viability of a second answer to my question: that humanoid forms might be required for the functions of some robots. Futurists have long imagined robotic carers in hospitals and day-care centres, looking after children, the elderly and the unwell. (Why robots are needed for these jobs is never quite explained.) Rather than otherworldly, eldritch contraptions, it’s thought humanoid robots are more appropriate in such human-centred situations. Unfortunately, humanoid robots need to make it through the ‘uncanny valley,’ a phenomenon wherein things that look almost human are far more unsettling than things that look rather less human. Think of bad computer graphics from the 90s, or any current attempts at expressive animatronic ‘human’ faces — not pleasant.

Unless the uncanny valley can be cleared, the utility of ‘lifelike’ humanoid robots is limited — it might actually be more effective to work with robots that have vaguely human forms (four limbs and a head), but don’t invoke feelings of fear or awkwardness. Think Big Hero 6, not The Polar Express.

But will society ever truly trust humanoid robots in the first place? Our fear of the not-quite-human permeates much of popular culture: hungry zombies, murderous androids and alien bodysnatchers have inspired countless books, film franchises and television series. The Cybermen in BBC’s long-running Doctor Who are almost pointlessly humanoid — just human brains in clumsy metal bodies — but their iconic form was dictated by poor special effects budgets in the 60s. Nevertheless, they've terrified generations of children. The misunderstood yet creepy replicants in Ridley Scott’s 1982 film Bladerunner are near-perfectly human except for their emotions; one wonders why such almost-people needed to be made in the first place. Couldn’t other types of robots have been just as effective in the fictional industrial workplaces in which they worked? It seems cruel to create creatures so dangerously close to humans when citizens fear them and their utility is questionable — this is something we should note down for future generations.

These sci-fi examples might say a lot about our relationship with humanoid robots — however, there may be some sense of societal inertia around their depiction: perhaps we want to create robots that look human because books, TV and movies keep telling us that we’re going to do so. Like a self-fulfilling prophecy, computer brains in person-shaped jars are being dragged into existence because we see no alternative future. But we shouldn’t feel beholden to Alex Garland’s Ex Machina or Steven Spielberg’s A.I. — other types of robots could serve our needs just as well, regardless of how sophisticated our own AI actually gets.

Just as human beings are not the centre of the Universe, neither should humanoid robots be the centre of our robotics efforts. Why bother? Honestly, it seems like a lot of hard work for nothing.

But robots still have a good amount of use, of course — that's why we devoted an entire issue to them! 

By Jack Scanlan

Jack is the Editor-in-Chief of Lateral.