The terrain is rough, autumn in New England, orange leaves thick on a steep hill. The creature struggles with a heavy pack, all four legs straining away, making its way upward, threading through the trees. At the plateau, someone gives it a swift kick in the side and it tumbles sideways, staggering, pivoting, compensating for the blow, one leg crossing back across the other, balletic, elegant. Balanced again, the creature would, to assess its surroundings, hold its head up — if it had a head. Instead there’s just a space where that should be, because the creature is a “BigDog” robot, symmetrical in either direction, designed to mimic the motion, balance and fluidity of an animal with the relentlessness of a machine.
The BigDog can move across just about any terrain, and it’s part of a family — the Cheetah, the LittleDog — designed in the spirit of robotics’ big idea of the millennium: that robots should neither be ever-more precise versions of blocky, mute machines, i.e. supercars, nor necessarily mimic the human shape, which has fairly specific applications and limits, but should instead use all the mechanics of nature, both imitated and recombined, to get past any number of technical impasses. The result, with something like the BigDog, is bizarre to behold, both beautiful and eerie, at once like an animal, vivomorphic, but also very different. When in the demonstration video a worker gives the machine a hard kick and it staggers, our emotions are inevitably engaged — such that one wanted to say it was kicked “harshly”, or “roughly”.
It is a stunning new insertion of technology into our lives, it is a triumph of the engineer’s art — and, of course, it is military-funded, developed under the Defense Advanced Research Projects Agency program. From those great folks who gave you the internet, a machine that could hunt you down and kill you across the breadth of a continent. Boston Dynamics, developed out of the tech web around MIT, blossomed under the massive investments of DARPA money — as did dozens of startups in the area — and now they’re coming home.
In 2013, Boston Dynamics, together with eight other corporations, was purchased by Google for an undisclosed sum, because why not? Why wouldn’t a search engine company want a robot army? It’s a measure of how far we’re into this new world that subsequent announcements from the Googleplex about new acquisitions — Acme Death Ray of Omaha, etc — barely raise a murmur. When Facebook announced six weeks ago that it was buying a drone start-up, there was bemusement, but of a very specific type. Why would Faceboook need a killer technology firm, the question went. It’s not as if they’re Google.
Killer technology — killer smart, killer new, killer killer. If the rise of 3D printing and additive manufacturing has brought forth a wave of utopian thinking — ways in which new technology could be distributed, democratised, self-printing printers turning out water filters and mobile med labs, busting the whole globalised consumer system with D-Ikea-Y — then the new robotics revolution appears to have brought out humanity’s other side. Both the imagination of what a new generation of robots might do and the way the money is flowing for their development is remorselessly in the direction of the military on the one hand, and for luxury service in the other. Killer robots from America, sex robots from Japan, personal majordomos to beat back drudgery, replace service workers, nurses and … who else? Who not?
The third side of the new robotics delta is thus a barely disguised fear of what is about to happen, or some hope will happen — the mass replacement of all sorts of repetitive labour by an unceasing cascade of new machines that will be first assembling themselves, and then designing themselves. Given that all this will be happening in a society that has, over recent decades, renounced most of the theories that would allow for a complex reorganisation of social life as a whole category of labour disappears and instead celebrates a market system that systematically excludes any capacity for reflection whatsoever, and you have a recipe for interesting times, i.e. a potential hell.
That something new is coming across the terrain — proprioceptive, biomimetic, indestructible — is without doubt. For the past decade, we have been caught up in a new robotics revolution, with the most impressive results visible to the general public only in the last five years or so. It’s true 3D printing is a technology that moves by first and starts, since so much of it involves the messiness of chemical processes, trying to find new ways to cook and form recalcitrant matter in the way we want it. Robotics is, for the most part, about something simpler: “robotics is about the control of motion in space”, one researcher/guru said to me, displaying a machine that removes tumours in a manner that, a few years ago, only gowned and deferred-to surgeons could. And once you realise the truth of that single statement, you can see why the field suddenly started to advance so fast, after decades of stasis. Robotics is essentially a series of very complex equations describing movement, put into action through metal and plastic.
Thus, the relentless advance in online shopping has been driven by the creation of “warehouse robots”, essentially a form of smart, moving storage that transforms a static warehouse into a dynamic unit; increasingly serious — but still routine — medical procedures such as blood-drawing have been robotised; the Alva 500 is a teleconference bot that not only patches you into meetings, but then drives itself to the next one you need to attend; a new wave of vision and steering technology has created major advances in the ability of robots to handle detailed terrain, including areas it has never been or had mapped into its memory; and a brilliantly simple new form of gripping, which replaces articulated pseudo-hands with a sort of “nerf ball” gripper which can handle objects of indefinite complexity.
None of these machines ascend to the level of the fantasy bot that fixes your drink before you get home and then repairs your car. But nor are they at the level of the Roomba, the forlorn little floor-vacuuming robot that came on the market a decade ago and seemed to symbolise all the disappointments of the first robot era (even though it was at the beginning of the second); like a Dalek, the Roomba scuttled across the floor but was easily defeated by stairs, surface changes or angry pets. The new generation of robots are not merely automating sequences of work, they are starting to replace jobs that required an ensemble and integration of skills.
That wasn’t happening 10 years ago. It wasn’t happening five years ago. It’s happening now, and it will start to get a lot faster. If you’re starting to feel that robots stories are popping up everywhere, get used to it; the spread will be a lot faster and more comprehensive than 3D printing.The question that suggests itself is not why now, but perhaps why did it take so long? Additive manufacturing and 3D printing are barely more than three decades old, whole new technologies invented well within the life-arc of an engineer still in mid-career. But robotics, as a dream, stretches back decades, centuries — further if you want to attach ideas of animated figures, golems and the like. Indeed, one could argue that robots — or the idea of them — precede the idea of flesh-and-blood workers, and make possible the leap of imagination that allows for the complex processes of human activity to be segmented into individual tasks. Adam Smith’s notorious celebration of the division of labour — whereby he uses the example of pin manufacture to urge the destruction of crafts and interesting and specialised work, in the interests of the general welfare — is essentially a prospectus for the human as robot. To that degree, the whole period of industrial, Taylorised labour is a prelude to the robot era.
“It might not be the million-dollar robot but the five-dollar machine that will change the world …”
The word — from the Czech robota, for “worker” — was coined in the ’20s. But it wasn’t until the 1960s that one- and then two-arm technology became possible and production lines began to be occupied by machines. Yet, though they began to be integrated into production — the “wow, it’s the future” science shows of the ’70s were filled with robotised Japanese production lines — the bright future of autonomous machines didn’t roll on, for a number of reasons. Lack of computing power was one; the portable computer didn’t exist, and vision-steered robotics was well beyond the processing capacity of of the time. A second issue was active resistance by workers in the West, who were understandably unwilling to acquiesce in their own replacement in gruelling but relatively well-paid low-skilled labour.
But the third reason that robots didn’t develop was because they weren’t needed — the rise of container shipping and the industrial development of Asian societies made mass-offshoring of production possible. Robots weren’t necessary because hundreds of millions of robotas were suddenly on the market, and the entire circuit of capital could be extended. When China opened up in the 1980s that supply appeared to be almost inexhaustible. The industrial robot market began to contract sharply from the mid-1980s onwards. For the 20 years that we think of as the period of “globalisation” — until the crash of 2008 — it’s as if we forgot about robots altogether. Their promise had initially been utopian, that we could enjoy a liberation from boring and dangerous labour hitherto available only to a few. When Asia opened, the argument was that it was exactly that boring labour that would be a liberation, promoting capitalist development. Much of the technological development was devoted to further integrating workers into ever-fasting production — such as in the notorious FoxConn factory in China, which took process automation to such a high pitch that it essentially regularised the production of suicide.
Robots might have disappeared from public view, but development of them continued in space and military applications. And by the 2000s, computing power — improving exponentially via Moore’s law — had passed the “midpoint” threshold, whereby the gains began to be increasingly impressive. By the early 2000s, computing power that in the 1980s had been confined to room-sized, gas-cooled Cray supercomputers designed to model nuclear explosions was now present in a mid-level gaming laptop. Putting such spatial-modelling technology into robots is what has allowed for the sudden animal-like leaps in the technology. Now the global race is on to robotise production — even China, with several hundred million potential workers still living in rural poverty, has committed itself to industrial robotisation (including the FoxConn factory); the impact on global society and economics is likely to be substantial.
But it will also be paradoxical, since the successive automation of production, while it benefits the individual firm, sector or country — to diminishing degrees — has the effect of reducing the circuit of capital. For the robotisation of production to be smoothly offset, significant new economic sectors must be created. That, throughout the history of capitalism, has been the promise of automation. There is every sign that this process stopped around the 1970s in the West, and a growing “surplus” population has been hidden by various statistical manoeuvres — and is now being actively persecuted and demonised.
What happens when the process of automation starts to not only accelerate, but to jerk forward — i.e. meta-accelerate, with the new industries and services designed to fill the employment gap, now pre-automated — and the innovations created therein feeding back new innovations in automation to the economy as a whole. Should that process occur, the relation of society to production changes — but in a way that we do not at the moment have the framework to accommodate in a calm and rational fashion. With an increasingly large, increasingly excluded population, how will dominant interests stay in control? Well, they’d need some pretty gruesome robots. Hmmmmmmmmm. If tonight, after watching the video, you dream of being chased by BigDog, it might well count as premonition.
That is the dark side of the new robotics. The more optimistic dimension comes from Moore’s law — computing capacity not only expands in power, but also falls in price. Robotics is no longer constrained within the increasingly monolithic military-information complex — the Google-mil-plex — but is increasingly being distributed.
It might not be the million-dollar robot but the five-dollar machine that will change the world — and those machines are coming from everywhere.
Related stories:
It’s hard not to imagine that robot picking itself up, turning slowly towards the kicker, and then extracting some sort of terrible relentless revenge.
Fascinating, thanks Guy.
@Frank – really? I get more the impression of a dog wondering why its human kicked it when all it wants to do is be the best dog it can be for its human.
Good article Guy – save the reference to Adam Smith. I wouldn’t quite say Smith ‘celebrated’ the division of labour, as he seemingly wavered on its social implications within the same text – deliberating upon the ‘mental mutilation’ such division can effect on its workers – although you rarely hear the provisos on any celebrated economic theory in general discourse.
I also think the jobs front is safe from robots, in general and in the near term – so long as companies still need people to have money in order to buy their products… The bigger worry is what these robots will enable their operators to be capable of and how ‘mutilated’ future jobs may be…
I recall reading a SF short story – it may have been Asimov -about a lunar war, where the enemy forces deployed small, rapidly moving multi-legged robots armed with explosives towards their objectives. They moved en masse, erratically, side to side, back and forth but inexorably towards their target. Terrifying.