Rise of the robots

But while many of DARPA’s projects elicit wry smiles from the media, the Special Weapons Observation Reconnaissance Detection System (SWORDS) has proved far more controversial. Introduced in 2004, the SWORDS robot is effectively a Talon with an automatic weapon strapped to its back and the ability to hit a bullseye from 2,000m away. In 2007, the SWORDS became the first weaponised robots to appear on a battlefield when three were dispatched to patrol roads in Iraq. The news quickly whipped up a storm of sci-fi-inspired panic, with wild reports claiming the SWORDS had targeted friendly soldiers and were being quietly pulled from service. The reports forced the robot’s creators, QinetiQ, to issue a hastily worded denial, but the furore refused to die down.

Now questions linger over the US military’s next step. In a book published by The National Academies Press in 2005, a Naval think-tank recommended that “the Navy and Marine Corps aggressively exploit the considerable war-fighting benefits offered by autonomous vehicles”. While far from a declaration of intent by the US Government itself, the suggestion did raise the unnerving prospect of fully automated battlefield robots deciding who to kill. Understandably, authorities have been wary of commenting on such reports but they do suggest the day is coming when armed robots outnumber humans on the battlefield.

However, the soldier closest to the SWORDS claims that robot-led wars are still out of the question. “You’ll never eliminate the soldier on the ground,” said Staff Sergeant Santiago Tordillos, the man who first demonstrated the SWORDS in action. “There’ll be a mix, but there will always be soldiers out there.” Though whether that means on the frontlines, or in support roles, he declined to elaborate.

Artificially unintelligent

Those already whistling the Terminator theme song under their breath can at least draw comfort from the opinions of academics, many of whom are sceptical regarding those 2025 claims. “There must be something in the air at the moment,” said Dr Steve Grand, one of Britain’s leading independent artificial-intelligence (AI) researchers. “A lot of people, including Microsoft, are pushing robotics as the ‘next big thing’ and I hope they’re right, but I’m cynical.”

He added: “We can make robots see, walk on two legs, manipulate objects, respond to commands, but only up to a point. Alan Turing said that by the end of the 20th century he expected us to talk routinely about machines being able to think for themselves. Yet when the year 2000 actually dawned, most of us were scared that our computers were too stupid to even add up the date properly. AI turned out to be much, much harder than anyone expected, mostly because we invariably underestimate the true difficulty of even the simplest of tasks faced by our brains.”

Grand argues that while robots such as Asimo can be created to fulfil specific tasks, the real challenge lies in creating a robot capable of adapting itself to its environment and taking on problems it hasn’t already been briefed to handle. In other words, creating a robot that isn’t just mimicking intelligence, but actually thinking for itself. Which is where Lucy comes into the equation. Lucy is a robotic orang-utan, which Grand hopes will one day achieve true intelligence by learning, one concept at a time, just as babies do. This sort of thinking is already driving a number of research projects in the US and Leipzig, where researchers are attempting to create a machine that teaches itself to walk one step at a time – rather than programming it with the instructions from birth. It’s a project applauded by Grand.