Pat Kane: Ghosting into the machine

Go into almost any high-street supermarket in Scotland today, and you will almost certainly be faced with a friendly neighbourhood robot. The self-operating checkout tills that have broken out like a rash over the last few years may not immediately seem like cousins of the killer androids from science-fiction.

But anyone who’s used them knows the robot moment. One absent-minded act – for example, passing an item from basket to bag without scanning it properly – and the screen notes that there is “an unscanned item in the bagging area”.

And won’t let you go any further until you remove it, and scan it again. Sometimes you – and the robot – get confused, requiring a real, live meat-machine to help you. But when you get up to speed with it, you marvel at its ability to sense exactly what item has been placed in your plastic bag.

My emotions are mixed before this machine. On one level, I can often enjoy the activity itself – there’s an element of self-help and dexterity about it. But on another level, I look over anxiously at the human-staffed tills. Should I be depriving some student, part-time mother or new immigrant the chance of a living wage? Or should I be happy that the relentless tide of automation is liberating workers from meaningless labour?

Economists and academics would say that any angst about the human-replacing effects of automation is historically faulty. Every innovation that’s enabled us to increase efficiency and productivity, to do more transformation of nature with less human labour, has led to an increase in the general sum of happiness – by making products cheaper, by forcing us to create jobs and sectors that turn out to create value in new and different ways.

That’s what humans are like with technology, they say. Look at the exponential rise that innovation has brought in living standards, longevity, material comfort. We can’t, and shouldn’t, be Luddites.

But what happens when these innovations start to simulate the very mental and skills-based agility that supposed to define us as human? When they can begin to understand language, and see the world as clearly as we do? When they have infinitely better memories, and can make connections and see patterns much more powerfully? We are already in an age where some the most unique human capacities for knowledge and decision are being gradually and casually automated.

It’s no news that chess grandmasters are regularly humbled by data-crunching mega-computers. Last year, the two human winners of the US general knowledge quiz show Jeopardy were trounced in live and direct competition with Watson, IBM’s latest silicon brain.

One of the most startling mental robotisations might even be the process of scientific discovery itself. A group at Cornell University has created a computer programme called Eureqa (available now, and for free, on the web). If you feed Eureqa enough data, it will formulate the kind of natural law that it took humans like Isaac Newton and William Hamilton to do. Last year, Eureqa was given information about the dynamics of a bacterium cell. After crunching through billions of equations, chucking away the irrelevant ones like failed species in evolution, it came to what its makers call a “beautiful, elegant equation that described how the cell worked, and that held true over all new experiments”.

The trouble was they had no idea how the machine had gotten there, or what underlying principle the equation expressed. “It was like consulting an oracle,” says Michael Schmidt, one of its makers, quoted in the online magazine Slate.

Schmidt thinks it’s possible that such computers may discover laws – derived from the unimaginably numerous interactions of genes, neurons and perhaps even markets – that we won’t be able to understand.

Eureqa, or its successors, will be in the position of “trying to explain Shakespeare to a dog”.

“Robot-assistance” is the friendlier face of automation. Surgeons are now finding that, in certain kinds of sutures and probings, a robot’s hand or instrument – though still guided by a doctor – is much more steady and precise than a human hand.

When looking for abnormalities in cervical smears, mammograms or coronary arteries, computers can now perform what’s known as “double reading” of scans – two doctors looking at the same results in order to better detect problems.

Hospitals are using much more autonomous robots, too. The new South Glasgow Hospitals Campus has ordered 22 to operate in a tunnel, tasked to distribute laundry, equipment, food and medicines. Yet we shouldn’t forget where the cutting-edge drive to make robots usable and ubiquitous in our lives comes from – and at least in Europe and America, that’s the military-industrial complex.

One company in Boston, iRobot, exemplifies the two-faces of robotics. It makes cute products like Roomba (a disc-like vacuum cleaner that navigates rooms autonomously) and Scooba (which does the same for bathroom floors). But the same tech also enables military robots like FirstLook, mobile video cameras which US soldiers can throw into windows before storming a building.

The US government’s Defense Advance Research and Projects Agency, or DARPA, funds the most wild-eyed robotics research. Look up the LS3 – Legged Squad Support System, being developed for the US Army – and be more than a little disturbed.

The LS3’s four legs can negotiate the roughest of terrain by itself, while its body can bear fuel and provisions for soldiers. Yet in its nervy, stuttering walk, it looks like a combination of a horse and a cockroach. One of its promotional videos shows the device cast upon a frozen lake, where its engineers kick and shove at it, trying but failing to overturn the machine.

A war horse, indeed – but usefully insensate to pain or abuse. A recent TED video showed yet another DARPA-supported project of tiny helicoptered drones, all chattering to each other and flying in formation – grids, figures of eight – and even buzzing out the James Bond theme tune.

The drones we know from the wars in Afghanistan are robots remotely guided by humans, thousands of miles away: the writer PW Singer quotes a young air force lieutenant who says “It’s like a video game with the ability to kill. It’s like ... freaking cool.”

Yet what the TED video shows is a spine-tingling degree of self-organisation among these machines. One of the creepier images is where the micro-drones swarm collectively and efficiently though a window, arranging themselves ready for action on the other side. These are robots as pack animals, or an insect swarm – at the fringes of your worst science-fiction nightmare.

But in another video, we see the same drones arranging themselves to steadily build a columnar construction out of rods, moving to a virtual architectural plan programmed into their tiny brains.

It doesn’t take too much projecting to see the near-future possibility of building sites requiring fewer workers, and much more machines of this precision, labouring tirelessly.

Undoubtedly, we need to exercise a public stake in the advance of automation. Applied purely under a commercial regime, robotisation leaves us with efficient factories, offices and industries, but hardly anyone with any wages left to buy the products and services they produce.

The joke among radicals these days is that it’s easier to imagine the end of the world than it is to imagine the end of capitalism. But I sometimes wonder if it’s even more difficult to imagine a perspective on automation which would subject it to the test of “benefitting the commonality”, as the original Luddites put it – rather than having us just tumble along in the wake of business, military and academic progress.

In short, I’d like to feel better about those wee women standing forlornly at the end of the supermarket check-out, watching their redundancy played out with every customer scan.

• Pat Kane’s forthcoming book is Radical Animal